early-access version 2820

README.md

@@ -1,7 +1,7 @@
 yuzu emulator early access
 =============
 
-This is the source code for early-access 2819.
+This is the source code for early-access 2820.
 
 ## Legal Notice
 

externals/cubeb/CMakeLists.txt

@@ -139,7 +139,7 @@ if(NOT BUNDLE_SPEEX)
 endif()
 
 if(NOT TARGET speex)
-  add_library(speex STATIC subprojects/speex/resample.c)
+  add_library(speex OBJECT subprojects/speex/resample.c)
   set_target_properties(speex PROPERTIES POSITION_INDEPENDENT_CODE TRUE)
   target_include_directories(speex INTERFACE subprojects)
   target_compile_definitions(speex PUBLIC
@@ -259,7 +259,7 @@ if(USE_WASAPI)
   target_sources(cubeb PRIVATE
     src/cubeb_wasapi.cpp)
   target_compile_definitions(cubeb PRIVATE USE_WASAPI)
-  target_link_libraries(cubeb PRIVATE avrt ole32)
+  target_link_libraries(cubeb PRIVATE avrt ole32 ksuser)
 endif()
 
 check_include_files("windows.h;mmsystem.h" USE_WINMM)
@@ -284,10 +284,23 @@ if(HAVE_SYS_SOUNDCARD_H)
   try_compile(USE_OSS "${PROJECT_BINARY_DIR}/compile_tests"
     ${PROJECT_SOURCE_DIR}/cmake/compile_tests/oss_is_v4.c)
   if(USE_OSS)
+    # strlcpy is not available on BSD systems that use glibc,
+    # like Debian kfreebsd, so try using libbsd if available
+    include(CheckSymbolExists)
+    check_symbol_exists(strlcpy string.h HAVE_STRLCPY)
+    if(NOT HAVE_STRLCPY)
+      pkg_check_modules(libbsd-overlay IMPORTED_TARGET libbsd-overlay)
+      if(libbsd-overlay_FOUND)
+        target_link_libraries(cubeb PRIVATE PkgConfig::libbsd-overlay)
+        set(HAVE_STRLCPY true)
+      endif()
+    endif()
+    if (HAVE_STRLCPY)
       target_sources(cubeb PRIVATE
         src/cubeb_oss.c)
       target_compile_definitions(cubeb PRIVATE USE_OSS)
     endif()
+  endif()
 endif()
 
 check_include_files(android/log.h USE_AUDIOTRACK)
@@ -367,7 +380,7 @@ if(BUILD_TESTS)
     add_executable(test_${NAME} test/test_${NAME}.cpp)
     target_include_directories(test_${NAME} PRIVATE ${gtest_SOURCE_DIR}/include src)
     target_link_libraries(test_${NAME} PRIVATE cubeb gtest_main)
-    add_test(${NAME} test_${NAME})
+    add_test(${NAME} test_${NAME} --gtest_death_test_style=threadsafe)
     add_sanitizers(test_${NAME})
     install(TARGETS test_${NAME})
   endmacro(cubeb_add_test)

externals/cubeb/INSTALL.md

@@ -2,7 +2,7 @@
 
 You must have CMake v3.1 or later installed.
 
-1. `git clone --recursive https://github.com/kinetiknz/cubeb.git`
+1. `git clone --recursive https://github.com/mozilla/cubeb.git`
 2. `mkdir cubeb-build`
 3. `cd cubeb-build`
 3. `cmake ../cubeb`

externals/cubeb/src/cubeb_aaudio.cpp

@@ -935,7 +935,8 @@ aaudio_stream_init_impl(cubeb_stream * stm, cubeb_devid input_device,
   stm->resampler = cubeb_resampler_create(
       stm, input_stream_params ? &in_params : NULL,
       output_stream_params ? &out_params : NULL, target_sample_rate,
-      stm->data_callback, stm->user_ptr, CUBEB_RESAMPLER_QUALITY_DEFAULT);
+      stm->data_callback, stm->user_ptr, CUBEB_RESAMPLER_QUALITY_DEFAULT,
+      CUBEB_RESAMPLER_RECLOCK_NONE);
 
   if (!stm->resampler) {
     LOG("Failed to create resampler");

externals/cubeb/src/cubeb_audiounit.cpp

@@ -1442,6 +1442,13 @@ audiounit_destroy(cubeb * ctx)
           audiounit_active_streams(ctx));
     }
 
+    // Destroying a cubeb context with device collection callbacks registered
+    // is misuse of the API, assert then attempt to clean up.
+    assert(!ctx->input_collection_changed_callback &&
+           !ctx->input_collection_changed_user_ptr &&
+           !ctx->output_collection_changed_callback &&
+           !ctx->output_collection_changed_user_ptr);
+
     /* Unregister the callback if necessary. */
     if (ctx->input_collection_changed_callback) {
       audiounit_remove_device_listener(ctx, CUBEB_DEVICE_TYPE_INPUT);
@@ -2700,7 +2707,8 @@ audiounit_setup_stream(cubeb_stream * stm)
   stm->resampler.reset(cubeb_resampler_create(
       stm, has_input(stm) ? &input_unconverted_params : NULL,
       has_output(stm) ? &stm->output_stream_params : NULL, target_sample_rate,
-      stm->data_callback, stm->user_ptr, CUBEB_RESAMPLER_QUALITY_DESKTOP));
+      stm->data_callback, stm->user_ptr, CUBEB_RESAMPLER_QUALITY_DESKTOP,
+      CUBEB_RESAMPLER_RECLOCK_NONE));
   if (!stm->resampler) {
     LOG("(%p) Could not create resampler.", stm);
     return CUBEB_ERROR;

externals/cubeb/src/cubeb_jack.cpp

@@ -925,15 +925,18 @@ cbjack_stream_init(cubeb * context, cubeb_stream ** stream,
   if (stm->devs == DUPLEX) {
     stm->resampler = cubeb_resampler_create(
         stm, &stm->in_params, &stm->out_params, stream_actual_rate,
-        stm->data_callback, stm->user_ptr, CUBEB_RESAMPLER_QUALITY_DESKTOP);
+        stm->data_callback, stm->user_ptr, CUBEB_RESAMPLER_QUALITY_DESKTOP,
+        CUBEB_RESAMPLER_RECLOCK_NONE);
   } else if (stm->devs == IN_ONLY) {
     stm->resampler = cubeb_resampler_create(
         stm, &stm->in_params, nullptr, stream_actual_rate, stm->data_callback,
-        stm->user_ptr, CUBEB_RESAMPLER_QUALITY_DESKTOP);
+        stm->user_ptr, CUBEB_RESAMPLER_QUALITY_DESKTOP,
+        CUBEB_RESAMPLER_RECLOCK_NONE);
   } else if (stm->devs == OUT_ONLY) {
     stm->resampler = cubeb_resampler_create(
         stm, nullptr, &stm->out_params, stream_actual_rate, stm->data_callback,
-        stm->user_ptr, CUBEB_RESAMPLER_QUALITY_DESKTOP);
+        stm->user_ptr, CUBEB_RESAMPLER_QUALITY_DESKTOP,
+        CUBEB_RESAMPLER_RECLOCK_NONE);
   }
 
   if (!stm->resampler) {

externals/cubeb/src/cubeb_opensl.c

@@ -1479,7 +1479,8 @@ opensl_stream_init(cubeb * ctx, cubeb_stream ** stream,
   stm->resampler = cubeb_resampler_create(
       stm, input_stream_params ? &input_params : NULL,
       output_stream_params ? &output_params : NULL, target_sample_rate,
-      data_callback, user_ptr, CUBEB_RESAMPLER_QUALITY_DEFAULT);
+      data_callback, user_ptr, CUBEB_RESAMPLER_QUALITY_DEFAULT,
+      CUBEB_RESAMPLER_RECLOCK_NONE);
   if (!stm->resampler) {
     LOG("Failed to create resampler");
     opensl_stream_destroy(stm);

externals/cubeb/src/cubeb_oss.c

@@ -96,9 +96,8 @@ struct oss_stream {
   oss_devnode_t name;
   int fd;
   void * buf;
-  unsigned int nfr; /* Number of frames allocated */
-  unsigned int nfrags;
   unsigned int bufframes;
+  unsigned int maxframes;
 
   struct stream_info {
     int channels;
@@ -824,9 +823,9 @@ retry:
       pfds[0].fd = s->play.fd;
       pfds[1].fd = -1;
       goto retry;
-    } else if (tnfr > (long)s->play.bufframes) {
+    } else if (tnfr > (long)s->play.maxframes) {
       /* too many frames available - limit */
-      tnfr = (long)s->play.bufframes;
+      tnfr = (long)s->play.maxframes;
     }
     if (nfr > tnfr) {
       nfr = tnfr;
@@ -842,9 +841,9 @@ retry:
       pfds[0].fd = -1;
       pfds[1].fd = s->record.fd;
       goto retry;
-    } else if (tnfr > (long)s->record.bufframes) {
+    } else if (tnfr > (long)s->record.maxframes) {
       /* too many frames available - limit */
-      tnfr = (long)s->record.bufframes;
+      tnfr = (long)s->record.maxframes;
     }
     if (nfr > tnfr) {
       nfr = tnfr;
@@ -1061,6 +1060,8 @@ oss_stream_init(cubeb * context, cubeb_stream ** stream,
   }
   if (input_stream_params != NULL) {
     unsigned int nb_channels;
+    uint32_t minframes;
+
     if (input_stream_params->prefs & CUBEB_STREAM_PREF_LOOPBACK) {
       LOG("Loopback not supported");
       ret = CUBEB_ERROR_NOT_SUPPORTED;
@@ -1089,18 +1090,17 @@ oss_stream_init(cubeb * context, cubeb_stream ** stream,
         (input_stream_params->format == CUBEB_SAMPLE_FLOAT32NE);
     s->record.frame_size =
         s->record.info.channels * (s->record.info.precision / 8);
-    s->record.nfrags = OSS_NFRAGS;
+    s->record.bufframes = latency_frames;
-    s->record.nfr = latency_frames / OSS_NFRAGS;
+
-    s->record.bufframes = s->record.nfrags * s->record.nfr;
+    oss_get_min_latency(context, *input_stream_params, &minframes);
-    uint32_t minnfr;
+    if (s->record.bufframes < minframes) {
-    oss_get_min_latency(context, *input_stream_params, &minnfr);
+      s->record.bufframes = minframes;
-    if (s->record.nfr < minnfr) {
-      s->record.nfr = minnfr;
-      s->record.nfrags = latency_frames / minnfr;
     }
   }
   if (output_stream_params != NULL) {
     unsigned int nb_channels;
+    uint32_t minframes;
+
     if (output_stream_params->prefs & CUBEB_STREAM_PREF_LOOPBACK) {
       LOG("Loopback not supported");
       ret = CUBEB_ERROR_NOT_SUPPORTED;
@@ -1128,19 +1128,16 @@ oss_stream_init(cubeb * context, cubeb_stream ** stream,
     }
     s->play.floating = (output_stream_params->format == CUBEB_SAMPLE_FLOAT32NE);
     s->play.frame_size = s->play.info.channels * (s->play.info.precision / 8);
-    s->play.nfrags = OSS_NFRAGS;
+    s->play.bufframes = latency_frames;
-    s->play.nfr = latency_frames / OSS_NFRAGS;
+
-    uint32_t minnfr;
+    oss_get_min_latency(context, *output_stream_params, &minframes);
-    oss_get_min_latency(context, *output_stream_params, &minnfr);
+    if (s->play.bufframes < minframes) {
-    if (s->play.nfr < minnfr) {
+      s->play.bufframes = minframes;
-      s->play.nfr = minnfr;
-      s->play.nfrags = latency_frames / minnfr;
     }
-    s->play.bufframes = s->play.nfrags * s->play.nfr;
   }
   if (s->play.fd != -1) {
     int frag = oss_get_frag_params(
-        oss_calc_frag_shift(s->play.nfr, s->play.frame_size));
+        oss_calc_frag_shift(s->play.bufframes, s->play.frame_size));
     if (ioctl(s->play.fd, SNDCTL_DSP_SETFRAGMENT, &frag))
       LOG("Failed to set play fd with SNDCTL_DSP_SETFRAGMENT. frag: 0x%x",
           frag);
@@ -1148,19 +1145,28 @@ oss_stream_init(cubeb * context, cubeb_stream ** stream,
     if (ioctl(s->play.fd, SNDCTL_DSP_GETOSPACE, &bi))
       LOG("Failed to get play fd's buffer info.");
     else {
-      s->play.nfr = bi.fragsize / s->play.frame_size;
+      s->play.bufframes = (bi.fragsize * bi.fragstotal) / s->play.frame_size;
-      s->play.nfrags = bi.fragments;
-      s->play.bufframes = s->play.nfr * s->play.nfrags;
     }
+    int lw;
 
-    int lw = s->play.frame_size;
+    /*
+     * Force 32 ms service intervals at most, or when recording is
+     * active, use the recording service intervals as a reference.
+     */
+    s->play.maxframes = (32 * output_stream_params->rate) / 1000;
+    if (s->record.fd != -1 || s->play.maxframes >= s->play.bufframes) {
+      lw = s->play.frame_size; /* Feed data when possible. */
+      s->play.maxframes = s->play.bufframes;
+    } else {
+      lw = (s->play.bufframes - s->play.maxframes) * s->play.frame_size;
+    }
     if (ioctl(s->play.fd, SNDCTL_DSP_LOW_WATER, &lw))
       LOG("Audio device \"%s\" (play) could not set trigger threshold",
           s->play.name);
   }
   if (s->record.fd != -1) {
     int frag = oss_get_frag_params(
-        oss_calc_frag_shift(s->record.nfr, s->record.frame_size));
+        oss_calc_frag_shift(s->record.bufframes, s->record.frame_size));
     if (ioctl(s->record.fd, SNDCTL_DSP_SETFRAGMENT, &frag))
       LOG("Failed to set record fd with SNDCTL_DSP_SETFRAGMENT. frag: 0x%x",
           frag);
@@ -1168,11 +1174,11 @@ oss_stream_init(cubeb * context, cubeb_stream ** stream,
     if (ioctl(s->record.fd, SNDCTL_DSP_GETISPACE, &bi))
       LOG("Failed to get record fd's buffer info.");
     else {
-      s->record.nfr = bi.fragsize / s->record.frame_size;
+      s->record.bufframes =
-      s->record.nfrags = bi.fragments;
+          (bi.fragsize * bi.fragstotal) / s->record.frame_size;
-      s->record.bufframes = s->record.nfr * s->record.nfrags;
     }
 
+    s->record.maxframes = s->record.bufframes;
     int lw = s->record.frame_size;
     if (ioctl(s->record.fd, SNDCTL_DSP_LOW_WATER, &lw))
       LOG("Audio device \"%s\" (record) could not set trigger threshold",

externals/cubeb/src/cubeb_pulse.c

@@ -783,6 +783,10 @@ pulse_context_destroy(cubeb * ctx)
 static void
 pulse_destroy(cubeb * ctx)
 {
+  assert(!ctx->input_collection_changed_callback &&
+         !ctx->input_collection_changed_user_ptr &&
+         !ctx->output_collection_changed_callback &&
+         !ctx->output_collection_changed_user_ptr);
   free(ctx->context_name);
   if (ctx->context) {
     pulse_context_destroy(ctx);

externals/cubeb/src/cubeb_resampler.cpp

@@ -323,7 +323,8 @@ cubeb_resampler_create(cubeb_stream * stream,
                        cubeb_stream_params * input_params,
                        cubeb_stream_params * output_params,
                        unsigned int target_rate, cubeb_data_callback callback,
-                       void * user_ptr, cubeb_resampler_quality quality)
+                       void * user_ptr, cubeb_resampler_quality quality,
+                       cubeb_resampler_reclock reclock)
 {
   cubeb_sample_format format;
 
@@ -337,13 +338,13 @@ cubeb_resampler_create(cubeb_stream * stream,
 
   switch (format) {
   case CUBEB_SAMPLE_S16NE:
-    return cubeb_resampler_create_internal<short>(stream, input_params,
+    return cubeb_resampler_create_internal<short>(
-                                                  output_params, target_rate,
+        stream, input_params, output_params, target_rate, callback, user_ptr,
-                                                  callback, user_ptr, quality);
+        quality, reclock);
   case CUBEB_SAMPLE_FLOAT32NE:
-    return cubeb_resampler_create_internal<float>(stream, input_params,
+    return cubeb_resampler_create_internal<float>(
-                                                  output_params, target_rate,
+        stream, input_params, output_params, target_rate, callback, user_ptr,
-                                                  callback, user_ptr, quality);
+        quality, reclock);
   default:
     assert(false);
     return nullptr;

externals/cubeb/src/cubeb_resampler.h

@@ -21,6 +21,11 @@ typedef enum {
   CUBEB_RESAMPLER_QUALITY_DESKTOP
 } cubeb_resampler_quality;
 
+typedef enum {
+  CUBEB_RESAMPLER_RECLOCK_NONE,
+  CUBEB_RESAMPLER_RECLOCK_INPUT
+} cubeb_resampler_reclock;
+
 /**
  * Create a resampler to adapt the requested sample rate into something that
  * is accepted by the audio backend.
@@ -44,7 +49,8 @@ cubeb_resampler_create(cubeb_stream * stream,
                        cubeb_stream_params * input_params,
                        cubeb_stream_params * output_params,
                        unsigned int target_rate, cubeb_data_callback callback,
-                       void * user_ptr, cubeb_resampler_quality quality);
+                       void * user_ptr, cubeb_resampler_quality quality,
+                       cubeb_resampler_reclock reclock);
 
 /**
  * Fill the buffer with frames acquired using the data callback. Resampling will

externals/cubeb/src/cubeb_resampler_internal.h

@@ -496,7 +496,8 @@ cubeb_resampler_create_internal(cubeb_stream * stream,
                                 cubeb_stream_params * output_params,
                                 unsigned int target_rate,
                                 cubeb_data_callback callback, void * user_ptr,
-                                cubeb_resampler_quality quality)
+                                cubeb_resampler_quality quality,
+                                cubeb_resampler_reclock reclock)
 {
   std::unique_ptr<cubeb_resampler_speex_one_way<T>> input_resampler = nullptr;
   std::unique_ptr<cubeb_resampler_speex_one_way<T>> output_resampler = nullptr;

externals/cubeb/src/cubeb_utils_win.h

@@ -11,24 +11,23 @@
 #include "cubeb-internal.h"
 #include <windows.h>
 
-/* This wraps a critical section to track the owner in debug mode, adapted from
+/* This wraps an SRWLock to track the owner in debug mode, adapted from
    NSPR and http://blogs.msdn.com/b/oldnewthing/archive/2013/07/12/10433554.aspx
  */
 class owned_critical_section {
 public:
   owned_critical_section()
+      : srwlock(SRWLOCK_INIT)
 #ifndef NDEBUG
-      : owner(0)
+        ,
+        owner(0)
 #endif
   {
-    InitializeCriticalSection(&critical_section);
   }
 
-  ~owned_critical_section() { DeleteCriticalSection(&critical_section); }
-
   void lock()
   {
-    EnterCriticalSection(&critical_section);
+    AcquireSRWLockExclusive(&srwlock);
 #ifndef NDEBUG
     XASSERT(owner != GetCurrentThreadId() && "recursive locking");
     owner = GetCurrentThreadId();
@@ -41,7 +40,7 @@ public:
     /* GetCurrentThreadId cannot return 0: it is not a the valid thread id */
     owner = 0;
 #endif
-    LeaveCriticalSection(&critical_section);
+    ReleaseSRWLockExclusive(&srwlock);
   }
 
   /* This is guaranteed to have the good behaviour if it succeeds. The behaviour
@@ -55,12 +54,12 @@ public:
   }
 
 private:
-  CRITICAL_SECTION critical_section;
+  SRWLOCK srwlock;
 #ifndef NDEBUG
   DWORD owner;
 #endif
 
-  // Disallow copy and assignment because CRICICAL_SECTION cannot be copied.
+  // Disallow copy and assignment because SRWLock cannot be copied.
   owned_critical_section(const owned_critical_section &);
   owned_critical_section & operator=(const owned_critical_section &);
 };

externals/cubeb/src/cubeb_wasapi.cpp

@@ -183,6 +183,46 @@ private:
 
 extern cubeb_ops const wasapi_ops;
 
+static com_heap_ptr<wchar_t>
+wasapi_get_default_device_id(EDataFlow flow, ERole role,
+                             IMMDeviceEnumerator * enumerator);
+
+struct wasapi_default_devices {
+  wasapi_default_devices(IMMDeviceEnumerator * enumerator)
+      : render_console_id(
+            wasapi_get_default_device_id(eRender, eConsole, enumerator)),
+        render_comms_id(
+            wasapi_get_default_device_id(eRender, eCommunications, enumerator)),
+        capture_console_id(
+            wasapi_get_default_device_id(eCapture, eConsole, enumerator)),
+        capture_comms_id(
+            wasapi_get_default_device_id(eCapture, eCommunications, enumerator))
+  {
+  }
+
+  bool is_default(EDataFlow flow, ERole role, wchar_t const * id)
+  {
+    wchar_t const * default_id = nullptr;
+    if (flow == eRender && role == eConsole) {
+      default_id = this->render_console_id.get();
+    } else if (flow == eRender && role == eCommunications) {
+      default_id = this->render_comms_id.get();
+    } else if (flow == eCapture && role == eConsole) {
+      default_id = this->capture_console_id.get();
+    } else if (flow == eCapture && role == eCommunications) {
+      default_id = this->capture_comms_id.get();
+    }
+
+    return default_id && wcscmp(id, default_id) == 0;
+  }
+
+private:
+  com_heap_ptr<wchar_t> render_console_id;
+  com_heap_ptr<wchar_t> render_comms_id;
+  com_heap_ptr<wchar_t> capture_console_id;
+  com_heap_ptr<wchar_t> capture_comms_id;
+};
+
 int
 wasapi_stream_stop(cubeb_stream * stm);
 int
@@ -195,7 +235,8 @@ ERole
 pref_to_role(cubeb_stream_prefs param);
 int
 wasapi_create_device(cubeb * ctx, cubeb_device_info & ret,
-                     IMMDeviceEnumerator * enumerator, IMMDevice * dev);
+                     IMMDeviceEnumerator * enumerator, IMMDevice * dev,
+                     wasapi_default_devices * defaults);
 void
 wasapi_destroy_device(cubeb_device_info * device_info);
 static int
@@ -216,6 +257,7 @@ class monitor_device_notifications;
 
 struct cubeb {
   cubeb_ops const * ops = &wasapi_ops;
+  owned_critical_section lock;
   cubeb_strings * device_ids;
   /* Device enumerator to get notifications when the
      device collection change. */
@@ -716,6 +758,8 @@ intern_device_id(cubeb * ctx, wchar_t const * id)
 {
   XASSERT(id);
 
+  auto_lock lock(ctx->lock);
+
   char const * tmp = wstr_to_utf8(id);
   if (!tmp) {
     return nullptr;
@@ -902,16 +946,16 @@ refill(cubeb_stream * stm, void * input_buffer, long input_frames_count,
   return out_frames;
 }
 
-int
+bool
 trigger_async_reconfigure(cubeb_stream * stm)
 {
   XASSERT(stm && stm->reconfigure_event);
+  LOG("Try reconfiguring the stream");
   BOOL ok = SetEvent(stm->reconfigure_event);
   if (!ok) {
     LOG("SetEvent on reconfigure_event failed: %lx", GetLastError());
-    return CUBEB_ERROR;
   }
-  return CUBEB_OK;
+  return static_cast<bool>(ok);
 }
 
 /* This helper grabs all the frames available from a capture client, put them in
@@ -940,8 +984,16 @@ get_input_buffer(cubeb_stream * stm)
     if (hr == AUDCLNT_E_DEVICE_INVALIDATED) {
       // Application can recover from this error. More info
       // https://msdn.microsoft.com/en-us/library/windows/desktop/dd316605(v=vs.85).aspx
-      LOG("Device invalidated error, reset default device");
+      LOG("Input device invalidated error");
-      trigger_async_reconfigure(stm);
+      // No need to reset device if user asks to use particular device, or
+      // switching is disabled.
+      if (stm->input_device_id ||
+          (stm->input_stream_params.prefs &
+           CUBEB_STREAM_PREF_DISABLE_DEVICE_SWITCHING) ||
+          !trigger_async_reconfigure(stm)) {
+        stm->state_callback(stm, stm->user_ptr, CUBEB_STATE_ERROR);
+        return false;
+      }
       return true;
     }
 
@@ -1046,8 +1098,16 @@ get_output_buffer(cubeb_stream * stm, void *& buffer, size_t & frame_count)
   if (hr == AUDCLNT_E_DEVICE_INVALIDATED) {
     // Application can recover from this error. More info
     // https://msdn.microsoft.com/en-us/library/windows/desktop/dd316605(v=vs.85).aspx
-    LOG("Device invalidated error, reset default device");
+    LOG("Output device invalidated error");
-    trigger_async_reconfigure(stm);
+    // No need to reset device if user asks to use particular device, or
+    // switching is disabled.
+    if (stm->output_device_id ||
+        (stm->output_stream_params.prefs &
+         CUBEB_STREAM_PREF_DISABLE_DEVICE_SWITCHING) ||
+        !trigger_async_reconfigure(stm)) {
+      stm->state_callback(stm, stm->user_ptr, CUBEB_STATE_ERROR);
+      return false;
+    }
     return true;
   }
 
@@ -1397,7 +1457,7 @@ wasapi_destroy(cubeb * context);
 HRESULT
 register_notification_client(cubeb_stream * stm)
 {
-  XASSERT(stm->device_enumerator);
+  XASSERT(stm->device_enumerator && !stm->notification_client);
 
   stm->notification_client.reset(new wasapi_endpoint_notification_client(
       stm->reconfigure_event, stm->role));
@@ -1415,7 +1475,7 @@ register_notification_client(cubeb_stream * stm)
 HRESULT
 unregister_notification_client(cubeb_stream * stm)
 {
-  XASSERT(stm->device_enumerator);
+  XASSERT(stm->device_enumerator && stm->notification_client);
 
   HRESULT hr = stm->device_enumerator->UnregisterEndpointNotificationCallback(
       stm->notification_client.get());
@@ -1454,6 +1514,9 @@ get_endpoint(com_ptr<IMMDevice> & device, LPCWSTR devid)
 HRESULT
 register_collection_notification_client(cubeb * context)
 {
+  context->lock.assert_current_thread_owns();
+  XASSERT(!context->device_collection_enumerator &&
+          !context->collection_notification_client);
   HRESULT hr = CoCreateInstance(
       __uuidof(MMDeviceEnumerator), NULL, CLSCTX_INPROC_SERVER,
       IID_PPV_ARGS(context->device_collection_enumerator.receive()));
@@ -1480,6 +1543,9 @@ register_collection_notification_client(cubeb * context)
 HRESULT
 unregister_collection_notification_client(cubeb * context)
 {
+  context->lock.assert_current_thread_owns();
+  XASSERT(context->device_collection_enumerator &&
+          context->collection_notification_client);
   HRESULT hr = context->device_collection_enumerator
                    ->UnregisterEndpointNotificationCallback(
                        context->collection_notification_client.get());
@@ -1608,6 +1674,7 @@ wasapi_init(cubeb ** context, char const * context_name)
   cubeb * ctx = new cubeb();
 
   ctx->ops = &wasapi_ops;
+  auto_lock lock(ctx->lock);
   if (cubeb_strings_init(&ctx->device_ids) != CUBEB_OK) {
     delete ctx;
     return CUBEB_ERROR;
@@ -1682,6 +1749,10 @@ stop_and_join_render_thread(cubeb_stream * stm)
 void
 wasapi_destroy(cubeb * context)
 {
+  auto_lock lock(context->lock);
+  XASSERT(!context->device_collection_enumerator &&
+          !context->collection_notification_client);
+
   if (context->device_ids) {
     cubeb_strings_destroy(context->device_ids);
   }
@@ -1887,7 +1958,7 @@ handle_channel_layout(cubeb_stream * stm, EDataFlow direction,
   }
 }
 
-static bool
+static int
 initialize_iaudioclient2(com_ptr<IAudioClient> & audio_client)
 {
   com_ptr<IAudioClient2> audio_client2;
@@ -1910,8 +1981,8 @@ initialize_iaudioclient2(com_ptr<IAudioClient> & audio_client)
   return CUBEB_OK;
 }
 
-// Not static to suppress a warning.
+#if 0
-/* static */ bool
+bool
 initialize_iaudioclient3(com_ptr<IAudioClient> & audio_client,
                          cubeb_stream * stm,
                          const com_heap_ptr<WAVEFORMATEX> & mix_format,
@@ -2021,6 +2092,7 @@ initialize_iaudioclient3(com_ptr<IAudioClient> & audio_client,
   LOG("Could not initialize shared stream with IAudioClient3: error: %lx", hr);
   return false;
 }
+#endif
 
 #define DIRECTION_NAME (direction == eCapture ? "capture" : "render")
 
@@ -2035,6 +2107,8 @@ setup_wasapi_stream_one_side(cubeb_stream * stm,
                              cubeb_stream_params * mix_params,
                              com_ptr<IMMDevice> & device)
 {
+  XASSERT(direction == eCapture || direction == eRender);
+
   HRESULT hr;
   bool is_loopback = stream_params->prefs & CUBEB_STREAM_PREF_LOOPBACK;
   if (is_loopback && direction != eCapture) {
@@ -2043,6 +2117,10 @@ setup_wasapi_stream_one_side(cubeb_stream * stm,
   }
 
   stm->stream_reset_lock.assert_current_thread_owns();
+  // If user doesn't specify a particular device, we can choose another one when
+  // the given devid is unavailable.
+  bool allow_fallback =
+      direction == eCapture ? !stm->input_device_id : !stm->output_device_id;
   bool try_again = false;
   // This loops until we find a device that works, or we've exhausted all
   // possibilities.
@@ -2092,7 +2170,7 @@ setup_wasapi_stream_one_side(cubeb_stream * stm,
           DIRECTION_NAME, hr);
       // A particular device can't be activated because it has been
       // unplugged, try fall back to the default audio device.
-      if (devid && hr == AUDCLNT_E_DEVICE_INVALIDATED) {
+      if (devid && hr == AUDCLNT_E_DEVICE_INVALIDATED && allow_fallback) {
         LOG("Trying again with the default %s audio device.", DIRECTION_NAME);
         devid = nullptr;
         device = nullptr;
@@ -2169,40 +2247,37 @@ setup_wasapi_stream_one_side(cubeb_stream * stm,
     flags |= AUDCLNT_STREAMFLAGS_EVENTCALLBACK;
   }
 
-  // Sanity check the latency, it may be that the device doesn't support it.
-  REFERENCE_TIME minimum_period;
-  REFERENCE_TIME default_period;
-  hr = audio_client->GetDevicePeriod(&default_period, &minimum_period);
-  if (FAILED(hr)) {
-    LOG("Could not get device period: %lx", hr);
-    return CUBEB_ERROR;
-  }
-
   REFERENCE_TIME latency_hns = frames_to_hns(stream_params->rate, stm->latency);
+
+  // Adjust input latency and check if input is using bluetooth handsfree
+  // protocol.
+  if (direction == eCapture) {
     stm->input_bluetooth_handsfree = false;
 
+    wasapi_default_devices default_devices(stm->device_enumerator.get());
     cubeb_device_info device_info;
     if (wasapi_create_device(stm->context, device_info,
-                           stm->device_enumerator.get(),
+                             stm->device_enumerator.get(), device.get(),
-                           device.get()) == CUBEB_OK) {
+                             &default_devices) == CUBEB_OK) {
+      // This multiplicator has been found empirically.
+      XASSERT(device_info.latency_hi > 0);
+      uint32_t latency_frames = device_info.latency_hi * 8;
+      LOG("Input: latency increased to %u frames from a default of %u",
+          latency_frames, device_info.latency_hi);
+      latency_hns = frames_to_hns(device_info.default_rate, latency_frames);
+
       const char * HANDSFREE_TAG = "BTHHFENUM";
       size_t len = sizeof(HANDSFREE_TAG);
-    if (direction == eCapture) {
-      uint32_t default_period_frames =
-          hns_to_frames(device_info.default_rate, default_period);
       if (strlen(device_info.group_id) >= len &&
           strncmp(device_info.group_id, HANDSFREE_TAG, len) == 0) {
+        LOG("Input device is using bluetooth handsfree protocol");
         stm->input_bluetooth_handsfree = true;
       }
-      // This multiplicator has been found empirically.
+
-      uint32_t latency_frames = default_period_frames * 8;
-      LOG("Input: latency increased to %u frames from a default of %u",
-          latency_frames, default_period_frames);
-      latency_hns = frames_to_hns(device_info.default_rate, latency_frames);
-    }
       wasapi_destroy_device(&device_info);
     } else {
-    LOG("Could not get cubeb_device_info.");
+      LOG("Could not get cubeb_device_info. Skip customizing input settings");
+    }
   }
 
   if (stream_params->prefs & CUBEB_STREAM_PREF_RAW) {
@@ -2258,8 +2333,10 @@ setup_wasapi_stream_one_side(cubeb_stream * stm,
 
 #undef DIRECTION_NAME
 
-void
+// Returns a non-null cubeb_devid if we find a matched device, or nullptr
-wasapi_find_matching_output_device(cubeb_stream * stm)
+// otherwise.
+cubeb_devid
+wasapi_find_bt_handsfree_output_device(cubeb_stream * stm)
 {
   HRESULT hr;
   cubeb_device_info * input_device = nullptr;
@@ -2268,19 +2345,21 @@ wasapi_find_matching_output_device(cubeb_stream * stm)
   // Only try to match to an output device if the input device is a bluetooth
   // device that is using the handsfree protocol
   if (!stm->input_bluetooth_handsfree) {
-    return;
+    return nullptr;
   }
 
   wchar_t * tmp = nullptr;
   hr = stm->input_device->GetId(&tmp);
   if (FAILED(hr)) {
-    LOG("Couldn't get input device id in wasapi_find_matching_output_device");
+    LOG("Couldn't get input device id in "
-    return;
+        "wasapi_find_bt_handsfree_output_device");
+    return nullptr;
   }
   com_heap_ptr<wchar_t> device_id(tmp);
-  cubeb_devid input_device_id = intern_device_id(stm->context, device_id.get());
+  cubeb_devid input_device_id = reinterpret_cast<cubeb_devid>(
+      intern_device_id(stm->context, device_id.get()));
   if (!input_device_id) {
-    return;
+    return nullptr;
   }
 
   int rv = wasapi_enumerate_devices(
@@ -2288,7 +2367,7 @@ wasapi_find_matching_output_device(cubeb_stream * stm)
       (cubeb_device_type)(CUBEB_DEVICE_TYPE_INPUT | CUBEB_DEVICE_TYPE_OUTPUT),
       &collection);
   if (rv != CUBEB_OK) {
-    return;
+    return nullptr;
   }
 
   // Find the input device, and then find the output device with the same group
@@ -2300,19 +2379,36 @@ wasapi_find_matching_output_device(cubeb_stream * stm)
     }
   }
 
+  cubeb_devid matched_output = nullptr;
+
+  if (input_device) {
     for (uint32_t i = 0; i < collection.count; i++) {
       cubeb_device_info & dev = collection.device[i];
-    if (dev.type == CUBEB_DEVICE_TYPE_OUTPUT && dev.group_id && input_device &&
+      if (dev.type == CUBEB_DEVICE_TYPE_OUTPUT && dev.group_id &&
           !strcmp(dev.group_id, input_device->group_id) &&
           dev.default_rate == input_device->default_rate) {
         LOG("Found matching device for %s: %s", input_device->friendly_name,
             dev.friendly_name);
-      stm->output_device_id =
+        matched_output = dev.devid;
-          utf8_to_wstr(reinterpret_cast<char const *>(dev.devid));
+        break;
+      }
     }
   }
 
   wasapi_device_collection_destroy(stm->context, &collection);
+  return matched_output;
+}
+
+std::unique_ptr<wchar_t[]>
+copy_wide_string(const wchar_t * src)
+{
+  XASSERT(src);
+  size_t len = wcslen(src);
+  std::unique_ptr<wchar_t[]> copy(new wchar_t[len + 1]);
+  if (wcsncpy_s(copy.get(), len + 1, src, len) != 0) {
+    return nullptr;
+  }
+  return copy;
 }
 
 int
@@ -2325,6 +2421,17 @@ setup_wasapi_stream(cubeb_stream * stm)
   XASSERT((!stm->output_client || !stm->input_client) &&
           "WASAPI stream already setup, close it first.");
 
+  std::unique_ptr<const wchar_t[]> selected_output_device_id;
+  if (stm->output_device_id) {
+    if (std::unique_ptr<wchar_t[]> tmp =
+            move(copy_wide_string(stm->output_device_id.get()))) {
+      selected_output_device_id = move(tmp);
+    } else {
+      LOG("Failed to copy output device identifier.");
+      return CUBEB_ERROR;
+    }
+  }
+
   if (has_input(stm)) {
     LOG("(%p) Setup capture: device=%p", stm, stm->input_device_id.get());
     rv = setup_wasapi_stream_one_side(
@@ -2356,8 +2463,12 @@ setup_wasapi_stream(cubeb_stream * stm)
     // device, and the default device is the same bluetooth device, pick the
     // right output device, running at the same rate and with the same protocol
     // as the input.
-    if (!stm->output_device_id) {
+    if (!selected_output_device_id) {
-      wasapi_find_matching_output_device(stm);
+      cubeb_devid matched = wasapi_find_bt_handsfree_output_device(stm);
+      if (matched) {
+        selected_output_device_id =
+            move(utf8_to_wstr(reinterpret_cast<char const *>(matched)));
+      }
     }
   }
 
@@ -2371,23 +2482,24 @@ setup_wasapi_stream(cubeb_stream * stm)
     stm->output_stream_params.channels = stm->input_stream_params.channels;
     stm->output_stream_params.layout = stm->input_stream_params.layout;
     if (stm->input_device_id) {
-      size_t len = wcslen(stm->input_device_id.get());
+      if (std::unique_ptr<wchar_t[]> tmp =
-      std::unique_ptr<wchar_t[]> tmp(new wchar_t[len + 1]);
+              move(copy_wide_string(stm->input_device_id.get()))) {
-      if (wcsncpy_s(tmp.get(), len + 1, stm->input_device_id.get(), len) != 0) {
+        XASSERT(!selected_output_device_id);
-        LOG("Failed to copy device identifier while copying input stream"
+        selected_output_device_id = move(tmp);
-            " configuration to output stream configuration to drive loopback.");
+      } else {
+        LOG("Failed to copy device identifier while copying input stream "
+            "configuration to output stream configuration to drive loopback.");
         return CUBEB_ERROR;
       }
-      stm->output_device_id = move(tmp);
     }
     stm->has_dummy_output = true;
   }
 
   if (has_output(stm)) {
-    LOG("(%p) Setup render: device=%p", stm, stm->output_device_id.get());
+    LOG("(%p) Setup render: device=%p", stm, selected_output_device_id.get());
     rv = setup_wasapi_stream_one_side(
-        stm, &stm->output_stream_params, stm->output_device_id.get(), eRender,
+        stm, &stm->output_stream_params, selected_output_device_id.get(),
-        __uuidof(IAudioRenderClient), stm->output_client,
+        eRender, __uuidof(IAudioRenderClient), stm->output_client,
         &stm->output_buffer_frame_count, stm->refill_event, stm->render_client,
         &stm->output_mix_params, stm->output_device);
     if (rv != CUBEB_OK) {
@@ -2448,10 +2560,11 @@ setup_wasapi_stream(cubeb_stream * stm)
 
   stm->resampler.reset(cubeb_resampler_create(
       stm, has_input(stm) ? &input_params : nullptr,
-      has_output(stm) ? &output_params : nullptr, target_sample_rate,
+      has_output(stm) && !stm->has_dummy_output ? &output_params : nullptr,
-      stm->data_callback, stm->user_ptr,
+      target_sample_rate, stm->data_callback, stm->user_ptr,
       stm->voice ? CUBEB_RESAMPLER_QUALITY_VOIP
-                 : CUBEB_RESAMPLER_QUALITY_DESKTOP));
+                 : CUBEB_RESAMPLER_QUALITY_DESKTOP,
+      CUBEB_RESAMPLER_RECLOCK_NONE));
   if (!stm->resampler) {
     LOG("Could not get a resampler");
     return CUBEB_ERROR;
@@ -2616,12 +2729,15 @@ wasapi_stream_init(cubeb * context, cubeb_stream ** stream,
     return rv;
   }
 
-  if (!((input_stream_params ? (input_stream_params->prefs &
+  // Follow the system default devices when not specifying devices explicitly
-                                CUBEB_STREAM_PREF_DISABLE_DEVICE_SWITCHING)
+  // and CUBEB_STREAM_PREF_DISABLE_DEVICE_SWITCHING is not set.
-                             : 0) ||
+  if ((!input_device && input_stream_params &&
-        (output_stream_params ? (output_stream_params->prefs &
+       !(input_stream_params->prefs &
-                                 CUBEB_STREAM_PREF_DISABLE_DEVICE_SWITCHING)
+         CUBEB_STREAM_PREF_DISABLE_DEVICE_SWITCHING)) ||
-                              : 0))) {
+      (!output_device && output_stream_params &&
+       !(output_stream_params->prefs &
+         CUBEB_STREAM_PREF_DISABLE_DEVICE_SWITCHING))) {
+    LOG("Follow the system default input or/and output devices");
     HRESULT hr = register_notification_client(stm.get());
     if (FAILED(hr)) {
       /* this is not fatal, we can still play audio, but we won't be able
@@ -3011,31 +3127,30 @@ static com_ptr<IMMDevice> wasapi_get_device_node(
   return ret;
 }
 
-static BOOL
+static com_heap_ptr<wchar_t>
-wasapi_is_default_device(EDataFlow flow, ERole role, LPCWSTR device_id,
+wasapi_get_default_device_id(EDataFlow flow, ERole role,
                              IMMDeviceEnumerator * enumerator)
 {
-  BOOL ret = FALSE;
   com_ptr<IMMDevice> dev;
-  HRESULT hr;
 
-  hr = enumerator->GetDefaultAudioEndpoint(flow, role, dev.receive());
+  HRESULT hr = enumerator->GetDefaultAudioEndpoint(flow, role, dev.receive());
   if (SUCCEEDED(hr)) {
     wchar_t * tmp = nullptr;
     if (SUCCEEDED(dev->GetId(&tmp))) {
-      com_heap_ptr<wchar_t> defdevid(tmp);
+      com_heap_ptr<wchar_t> devid(tmp);
-      ret = (wcscmp(defdevid.get(), device_id) == 0);
+      return devid;
     }
   }
 
-  return ret;
+  return nullptr;
 }
 
 /* `ret` must be deallocated with `wasapi_destroy_device`, iff the return value
  * of this function is `CUBEB_OK`. */
 int
 wasapi_create_device(cubeb * ctx, cubeb_device_info & ret,
-                     IMMDeviceEnumerator * enumerator, IMMDevice * dev)
+                     IMMDeviceEnumerator * enumerator, IMMDevice * dev,
+                     wasapi_default_devices * defaults)
 {
   com_ptr<IMMEndpoint> endpoint;
   com_ptr<IMMDevice> devnode;
@@ -3046,6 +3161,8 @@ wasapi_create_device(cubeb * ctx, cubeb_device_info & ret,
   REFERENCE_TIME def_period, min_period;
   HRESULT hr;
 
+  XASSERT(enumerator && dev && defaults);
+
   // zero-out to be able to safely delete the pointers to friendly_name and
   // group_id at all time in this function.
   PodZero(&ret, 1);
@@ -3133,19 +3250,14 @@ wasapi_create_device(cubeb * ctx, cubeb_device_info & ret,
   }
 
   ret.preferred = CUBEB_DEVICE_PREF_NONE;
-  if (wasapi_is_default_device(flow, eConsole, device_id.get(), enumerator)) {
+  if (defaults->is_default(flow, eConsole, device_id.get())) {
     ret.preferred =
-        (cubeb_device_pref)(ret.preferred | CUBEB_DEVICE_PREF_MULTIMEDIA);
+        (cubeb_device_pref)(ret.preferred | CUBEB_DEVICE_PREF_MULTIMEDIA |
-  }
+                            CUBEB_DEVICE_PREF_NOTIFICATION);
-  if (wasapi_is_default_device(flow, eCommunications, device_id.get(),
+  } else if (defaults->is_default(flow, eCommunications, device_id.get())) {
-                               enumerator)) {
     ret.preferred =
         (cubeb_device_pref)(ret.preferred | CUBEB_DEVICE_PREF_VOICE);
   }
-  if (wasapi_is_default_device(flow, eConsole, device_id.get(), enumerator)) {
-    ret.preferred =
-        (cubeb_device_pref)(ret.preferred | CUBEB_DEVICE_PREF_NOTIFICATION);
-  }
 
   if (flow == eRender) {
     ret.type = CUBEB_DEVICE_TYPE_OUTPUT;
@@ -3229,14 +3341,17 @@ wasapi_enumerate_devices(cubeb * context, cubeb_device_type type,
     return CUBEB_ERROR;
   }
 
-  if (type == CUBEB_DEVICE_TYPE_OUTPUT)
+  wasapi_default_devices default_devices(enumerator.get());
+
+  if (type == CUBEB_DEVICE_TYPE_OUTPUT) {
     flow = eRender;
-  else if (type == CUBEB_DEVICE_TYPE_INPUT)
+  } else if (type == CUBEB_DEVICE_TYPE_INPUT) {
     flow = eCapture;
-  else if (type & (CUBEB_DEVICE_TYPE_INPUT | CUBEB_DEVICE_TYPE_OUTPUT))
+  } else if (type & (CUBEB_DEVICE_TYPE_INPUT | CUBEB_DEVICE_TYPE_OUTPUT)) {
     flow = eAll;
-  else
+  } else {
     return CUBEB_ERROR;
+  }
 
   hr = enumerator->EnumAudioEndpoints(flow, DEVICE_STATEMASK_ALL,
                                       collection.receive());
@@ -3264,7 +3379,7 @@ wasapi_enumerate_devices(cubeb * context, cubeb_device_type type,
       continue;
     }
     if (wasapi_create_device(context, devices[out->count], enumerator.get(),
-                             dev.get()) == CUBEB_OK) {
+                             dev.get(), &default_devices) == CUBEB_OK) {
       out->count += 1;
     }
   }
@@ -3294,6 +3409,7 @@ wasapi_register_device_collection_changed(
     cubeb_device_collection_changed_callback collection_changed_callback,
     void * user_ptr)
 {
+  auto_lock lock(context->lock);
   if (devtype == CUBEB_DEVICE_TYPE_UNKNOWN) {
     return CUBEB_ERROR_INVALID_PARAMETER;
   }

externals/cubeb/src/cubeb_winmm.c

@@ -73,17 +73,6 @@
 #define CUBEB_STREAM_MAX 32
 #define NBUFS 4
 
-const GUID KSDATAFORMAT_SUBTYPE_PCM = {
-    0x00000001,
-    0x0000,
-    0x0010,
-    {0x80, 0x00, 0x00, 0xaa, 0x00, 0x38, 0x9b, 0x71}};
-const GUID KSDATAFORMAT_SUBTYPE_IEEE_FLOAT = {
-    0x00000003,
-    0x0000,
-    0x0010,
-    {0x80, 0x00, 0x00, 0xaa, 0x00, 0x38, 0x9b, 0x71}};
-
 struct cubeb_stream_item {
   SLIST_ENTRY head;
   cubeb_stream * stream;

externals/cubeb/test/test_duplex.cpp

@@ -137,26 +137,20 @@ void device_collection_changed_callback(cubeb * context, void * user)
                         " called when opening a stream";
 }
 
-TEST(cubeb, duplex_collection_change)
+void
+duplex_collection_change_impl(cubeb * ctx)
 {
-  cubeb *ctx;
+  cubeb_stream * stream;
-  cubeb_stream *stream;
   cubeb_stream_params input_params;
   cubeb_stream_params output_params;
   int r;
   uint32_t latency_frames = 0;
 
-  r = common_init(&ctx, "Cubeb duplex example with collection change");
+  r = cubeb_register_device_collection_changed(
-  ASSERT_EQ(r, CUBEB_OK) << "Error initializing cubeb library";
+      ctx, static_cast<cubeb_device_type>(CUBEB_DEVICE_TYPE_INPUT),
-
+      device_collection_changed_callback, nullptr);
-  r = cubeb_register_device_collection_changed(ctx,
-                                               static_cast<cubeb_device_type>(CUBEB_DEVICE_TYPE_INPUT),
-                                               device_collection_changed_callback,
-                                               nullptr);
   ASSERT_EQ(r, CUBEB_OK) << "Error initializing cubeb stream";
 
-  std::unique_ptr<cubeb, decltype(&cubeb_destroy)>
-    cleanup_cubeb_at_exit(ctx, cubeb_destroy);
 
   /* typical user-case: mono input, stereo output, low latency. */
   input_params.format = STREAM_FORMAT;
@@ -173,13 +167,43 @@ TEST(cubeb, duplex_collection_change)
   r = cubeb_get_min_latency(ctx, &output_params, &latency_frames);
   ASSERT_EQ(r, CUBEB_OK) << "Could not get minimal latency";
 
-  r = cubeb_stream_init(ctx, &stream, "Cubeb duplex",
+  r = cubeb_stream_init(ctx, &stream, "Cubeb duplex", NULL, &input_params, NULL,
-                        NULL, &input_params, NULL, &output_params,
+                        &output_params, latency_frames, data_cb_duplex,
-                        latency_frames, data_cb_duplex, state_cb_duplex, nullptr);
+                        state_cb_duplex, nullptr);
   ASSERT_EQ(r, CUBEB_OK) << "Error initializing cubeb stream";
   cubeb_stream_destroy(stream);
 }
 
+TEST(cubeb, duplex_collection_change)
+{
+  cubeb * ctx;
+  int r;
+
+  r = common_init(&ctx, "Cubeb duplex example with collection change");
+  ASSERT_EQ(r, CUBEB_OK) << "Error initializing cubeb library";
+  std::unique_ptr<cubeb, decltype(&cubeb_destroy)> cleanup_cubeb_at_exit(
+      ctx, cubeb_destroy);
+
+  duplex_collection_change_impl(ctx);
+  r = cubeb_register_device_collection_changed(
+      ctx, static_cast<cubeb_device_type>(CUBEB_DEVICE_TYPE_INPUT), nullptr,
+      nullptr);
+  ASSERT_EQ(r, CUBEB_OK);
+}
+
+TEST(cubeb, duplex_collection_change_no_unregister)
+{
+  cubeb * ctx;
+  int r;
+
+  r = common_init(&ctx, "Cubeb duplex example with collection change");
+  ASSERT_EQ(r, CUBEB_OK) << "Error initializing cubeb library";
+  std::unique_ptr<cubeb, decltype(&cubeb_destroy)> cleanup_cubeb_at_exit(
+      ctx, [](cubeb * p) noexcept { EXPECT_DEATH(cubeb_destroy(p), ""); });
+
+  duplex_collection_change_impl(ctx);
+}
+
 long data_cb_input(cubeb_stream * stream, void * user, const void * inputbuffer, void * outputbuffer, long nframes)
 {
   if (stream == NULL || inputbuffer == NULL || outputbuffer != NULL) {

externals/cubeb/test/test_resampler.cpp

@@ -338,7 +338,8 @@ void test_resampler_duplex(uint32_t input_channels, uint32_t output_channels,
 
   cubeb_resampler * resampler =
     cubeb_resampler_create((cubeb_stream*)nullptr, &input_params, &output_params, target_rate,
-                           data_cb_resampler, (void*)&state, CUBEB_RESAMPLER_QUALITY_VOIP);
+                           data_cb_resampler, (void*)&state, CUBEB_RESAMPLER_QUALITY_VOIP,
+                           CUBEB_RESAMPLER_RECLOCK_NONE);
 
   long latency = cubeb_resampler_latency(resampler);
 
@@ -484,8 +485,8 @@ TEST(cubeb, resampler_output_only_noop)
   cubeb_resampler * resampler =
     cubeb_resampler_create((cubeb_stream*)nullptr, nullptr, &output_params, target_rate,
                            test_output_only_noop_data_cb, nullptr,
-                           CUBEB_RESAMPLER_QUALITY_VOIP);
+                           CUBEB_RESAMPLER_QUALITY_VOIP,
-
+                           CUBEB_RESAMPLER_RECLOCK_NONE);
   const long out_frames = 128;
   float out_buffer[out_frames];
   long got;
@@ -523,7 +524,8 @@ TEST(cubeb, resampler_drain)
   cubeb_resampler * resampler =
     cubeb_resampler_create((cubeb_stream*)nullptr, nullptr, &output_params, target_rate,
                            test_drain_data_cb, &cb_count,
-                           CUBEB_RESAMPLER_QUALITY_VOIP);
+                           CUBEB_RESAMPLER_QUALITY_VOIP,
+                           CUBEB_RESAMPLER_RECLOCK_NONE);
 
   const long out_frames = 128;
   float out_buffer[out_frames];
@@ -572,7 +574,8 @@ TEST(cubeb, resampler_passthrough_output_only)
   cubeb_resampler * resampler =
     cubeb_resampler_create((cubeb_stream*)nullptr, nullptr, &output_params,
                            target_rate, cb_passthrough_resampler_output, nullptr,
-                           CUBEB_RESAMPLER_QUALITY_VOIP);
+                           CUBEB_RESAMPLER_QUALITY_VOIP,
+                           CUBEB_RESAMPLER_RECLOCK_NONE);
 
   float output_buffer[output_channels * 256];
 
@@ -616,7 +619,8 @@ TEST(cubeb, resampler_passthrough_input_only)
   cubeb_resampler * resampler =
     cubeb_resampler_create((cubeb_stream*)nullptr, &input_params, nullptr,
                            target_rate, cb_passthrough_resampler_input, nullptr,
-                           CUBEB_RESAMPLER_QUALITY_VOIP);
+                           CUBEB_RESAMPLER_QUALITY_VOIP,
+                           CUBEB_RESAMPLER_RECLOCK_NONE);
 
   float input_buffer[input_channels * 256];
 
@@ -737,7 +741,8 @@ TEST(cubeb, resampler_passthrough_duplex_callback_reordering)
   cubeb_resampler * resampler =
     cubeb_resampler_create((cubeb_stream*)nullptr, &input_params, &output_params,
                            target_rate, cb_passthrough_resampler_duplex, &c,
-                           CUBEB_RESAMPLER_QUALITY_VOIP);
+                           CUBEB_RESAMPLER_QUALITY_VOIP,
+                           CUBEB_RESAMPLER_RECLOCK_NONE);
 
   const long BUF_BASE_SIZE = 256;
   float input_buffer_prebuffer[input_channels * BUF_BASE_SIZE * 2];
@@ -820,7 +825,7 @@ TEST(cubeb, resampler_drift_drop_data)
     cubeb_resampler * resampler =
       cubeb_resampler_create((cubeb_stream*)nullptr, &input_params, &output_params,
         target_rate, cb_passthrough_resampler_duplex, &c,
-        CUBEB_RESAMPLER_QUALITY_VOIP);
+        CUBEB_RESAMPLER_QUALITY_VOIP, CUBEB_RESAMPLER_RECLOCK_NONE);
 
     const long BUF_BASE_SIZE = 256;
 

externals/cubeb/tools/cubeb-test.cpp

@@ -7,6 +7,7 @@
 #include <cstring>
 #include <inttypes.h>
 #include <iostream>
+#include <vector>
 #ifdef _WIN32
 #include <objbase.h> // Used by CoInitialize()
 #endif
@@ -35,6 +36,32 @@ static const char* state_to_string(cubeb_state state) {
   }
 }
 
+static const char* device_type_to_string(cubeb_device_type type) {
+  switch (type) {
+    case CUBEB_DEVICE_TYPE_INPUT:
+      return "input";
+    case CUBEB_DEVICE_TYPE_OUTPUT:
+      return "output";
+    case CUBEB_DEVICE_TYPE_UNKNOWN:
+      return "unknown";
+    default:
+      assert(false);
+  }
+}
+
+static const char* device_state_to_string(cubeb_device_state state) {
+  switch (state) {
+    case CUBEB_DEVICE_STATE_DISABLED:
+      return "disabled";
+    case CUBEB_DEVICE_STATE_UNPLUGGED:
+      return "unplugged";
+    case CUBEB_DEVICE_STATE_ENABLED:
+      return "enabled";
+    default:
+      assert(false);
+  }
+}
+
 void print_log(const char* msg, ...) {
   va_list args;
   va_start(args, msg);
@@ -48,6 +75,7 @@ public:
   ~cubeb_client() {}
 
   bool init(char const * backend_name = nullptr);
+  cubeb_devid select_device(cubeb_device_type type) const;
   bool init_stream();
   bool start_stream();
   bool stop_stream();
@@ -70,7 +98,10 @@ public:
   bool unregister_device_collection_changed(cubeb_device_type devtype) const;
 
   cubeb_stream_params output_params = {};
+  cubeb_devid output_device = nullptr;
+
   cubeb_stream_params input_params = {};
+  cubeb_devid input_device = nullptr;
 
   void force_drain() { _force_drain = true; }
 
@@ -81,8 +112,6 @@ private:
   cubeb* context = nullptr;
 
   cubeb_stream* stream = nullptr;
-  cubeb_devid output_device = nullptr;
-  cubeb_devid input_device = nullptr;
 
   /* Accessed only from client and audio thread. */
   std::atomic<uint32_t> _rate = {0};
@@ -492,6 +521,56 @@ bool choose_action(cubeb_client& cl, operation_data * op, int c) {
   return true; // Loop up
 }
 
+cubeb_devid cubeb_client::select_device(cubeb_device_type type) const
+{
+  assert(type == CUBEB_DEVICE_TYPE_INPUT || type == CUBEB_DEVICE_TYPE_OUTPUT);
+
+  cubeb_device_collection collection;
+  if (cubeb_enumerate_devices(context, type, &collection) ==
+      CUBEB_ERROR_NOT_SUPPORTED) {
+    fprintf(stderr,
+            "Not support %s device selection. Force to use default device\n",
+            device_type_to_string(type));
+    return nullptr;
+  }
+
+  assert(collection.count);
+  fprintf(stderr, "Found %zu %s devices. Choose one:\n", collection.count,
+          device_type_to_string(type));
+
+  std::vector<cubeb_devid> devices;
+  devices.emplace_back(nullptr);
+  fprintf(stderr, "# 0\n\tname: system default device\n");
+  for (size_t i = 0; i < collection.count; i++) {
+    assert(collection.device[i].type == type);
+    fprintf(stderr,
+            "# %zu %s\n"
+            "\tname: %s\n"
+            "\tdevice id: %s\n"
+            "\tmax channels: %u\n"
+            "\tstate: %s\n",
+            devices.size(),
+            collection.device[i].preferred ? " (PREFERRED)" : "",
+            collection.device[i].friendly_name, collection.device[i].device_id,
+            collection.device[i].max_channels,
+            device_state_to_string(collection.device[i].state));
+    devices.emplace_back(collection.device[i].devid);
+  }
+
+  cubeb_device_collection_destroy(context, &collection);
+
+  size_t number;
+  std::cout << "Enter device number: ";
+  std::cin >> number;
+  while (!std::cin || number >= devices.size()) {
+    std::cin.clear();
+    std::cin.ignore(100, '\n');
+    std::cout << "Error: Please enter a valid numeric input. Enter again: ";
+    std::cin >> number;
+  }
+  return devices[number];
+}
+
 int main(int argc, char* argv[]) {
 #ifdef _WIN32
   CoInitialize(nullptr);
@@ -523,7 +602,7 @@ int main(int argc, char* argv[]) {
 
   bool res = false;
   cubeb_client cl;
-  cl.activate_log(CUBEB_LOG_DISABLED);
+  cl.activate_log(CUBEB_LOG_NORMAL);
   fprintf(stderr, "Log level is DISABLED\n");
   cl.init(/* default backend */);
 
@@ -540,10 +619,12 @@ int main(int argc, char* argv[]) {
     }
   } else {
     if (op.pm == PLAYBACK || op.pm == DUPLEX || op.pm == LATENCY_TESTING) {
+      cl.output_device = cl.select_device(CUBEB_DEVICE_TYPE_OUTPUT);
       cl.output_params = {CUBEB_SAMPLE_FLOAT32NE, op.rate, DEFAULT_OUTPUT_CHANNELS,
                           CUBEB_LAYOUT_STEREO, CUBEB_STREAM_PREF_NONE};
     }
     if (op.pm == RECORD || op.pm == DUPLEX || op.pm == LATENCY_TESTING) {
+      cl.input_device = cl.select_device(CUBEB_DEVICE_TYPE_INPUT);
       cl.input_params = {CUBEB_SAMPLE_FLOAT32NE, op.rate, DEFAULT_INPUT_CHANNELS, CUBEB_LAYOUT_UNDEFINED, CUBEB_STREAM_PREF_NONE};
     }
     if (op.pm == LATENCY_TESTING) {

src/audio_core/CMakeLists.txt

@@ -1,54 +1,214 @@
 add_library(audio_core STATIC
-    algorithm/filter.cpp
+    audio_core.cpp
-    algorithm/filter.h
+    audio_core.h
-    algorithm/interpolate.cpp
+    audio_event.h
-    algorithm/interpolate.h
+    audio_event.cpp
-    audio_out.cpp
+    audio_render_manager.cpp
-    audio_out.h
+    audio_render_manager.h
-    audio_renderer.cpp
+    audio_in_manager.cpp
-    audio_renderer.h
+    audio_in_manager.h
-    behavior_info.cpp
+    audio_out_manager.cpp
-    behavior_info.h
+    audio_out_manager.h
-    buffer.h
+    audio_manager.cpp
-    codec.cpp
+    audio_manager.h
-    codec.h
+    common/audio_renderer_parameter.h
-    command_generator.cpp
+    common/common.h
-    command_generator.h
+    common/feature_support.h
-    common.h
+    common/wave_buffer.h
-    delay_line.cpp
+    common/workbuffer_allocator.h
-    delay_line.h
+    device/audio_buffer.h
-    effect_context.cpp
+    device/audio_buffers.h
-    effect_context.h
+    device/device_session.cpp
-    info_updater.cpp
+    device/device_session.h
-    info_updater.h
+    in/audio_in.cpp
-    memory_pool.cpp
+    in/audio_in.h
-    memory_pool.h
+    in/audio_in_system.cpp
-    mix_context.cpp
+    in/audio_in_system.h
-    mix_context.h
+    out/audio_out.cpp
-    null_sink.h
+    out/audio_out.h
-    sink.h
+    out/audio_out_system.cpp
-    sink_context.cpp
+    out/audio_out_system.h
-    sink_context.h
+    renderer/adsp/adsp.cpp
-    sink_details.cpp
+    renderer/adsp/adsp.h
-    sink_details.h
+    renderer/adsp/audio_renderer.cpp
-    sink_stream.h
+    renderer/adsp/audio_renderer.h
-    splitter_context.cpp
+    renderer/adsp/command_buffer.h
-    splitter_context.h
+    renderer/adsp/command_list_processor.cpp
-    stream.cpp
+    renderer/adsp/command_list_processor.h
-    stream.h
+    renderer/audio_device.cpp
-    voice_context.cpp
+    renderer/audio_device.h
-    voice_context.h
+    renderer/audio_renderer.h
-
+    renderer/audio_renderer.cpp
-    $<$<BOOL:${ENABLE_CUBEB}>:cubeb_sink.cpp cubeb_sink.h>
+    renderer/behavior/behavior_info.cpp
-    $<$<BOOL:${ENABLE_SDL2}>:sdl2_sink.cpp sdl2_sink.h>
+    renderer/behavior/behavior_info.h
+    renderer/behavior/info_updater.cpp
+    renderer/behavior/info_updater.h
+    renderer/command/data_source/adpcm.cpp
+    renderer/command/data_source/adpcm.h
+    renderer/command/data_source/decode.cpp
+    renderer/command/data_source/decode.h
+    renderer/command/data_source/pcm_float.cpp
+    renderer/command/data_source/pcm_float.h
+    renderer/command/data_source/pcm_int16.cpp
+    renderer/command/data_source/pcm_int16.h
+    renderer/command/effect/aux_.cpp
+    renderer/command/effect/aux_.h
+    renderer/command/effect/biquad_filter.cpp
+    renderer/command/effect/biquad_filter.h
+    renderer/command/effect/capture.cpp
+    renderer/command/effect/capture.h
+    renderer/command/effect/delay.cpp
+    renderer/command/effect/delay.h
+    renderer/command/effect/i3dl2_reverb.cpp
+    renderer/command/effect/i3dl2_reverb.h
+    renderer/command/effect/light_limiter.cpp
+    renderer/command/effect/light_limiter.h
+    renderer/command/effect/multi_tap_biquad_filter.cpp
+    renderer/command/effect/multi_tap_biquad_filter.h
+    renderer/command/effect/reverb.cpp
+    renderer/command/effect/reverb.h
+    renderer/command/mix/clear_mix.cpp
+    renderer/command/mix/clear_mix.h
+    renderer/command/mix/copy_mix.cpp
+    renderer/command/mix/copy_mix.h
+    renderer/command/mix/depop_for_mix_buffers.cpp
+    renderer/command/mix/depop_for_mix_buffers.h
+    renderer/command/mix/depop_prepare.cpp
+    renderer/command/mix/depop_prepare.h
+    renderer/command/mix/mix.cpp
+    renderer/command/mix/mix.h
+    renderer/command/mix/mix_ramp.cpp
+    renderer/command/mix/mix_ramp.h
+    renderer/command/mix/mix_ramp_grouped.cpp
+    renderer/command/mix/mix_ramp_grouped.h
+    renderer/command/mix/volume.cpp
+    renderer/command/mix/volume.h
+    renderer/command/mix/volume_ramp.cpp
+    renderer/command/mix/volume_ramp.h
+    renderer/command/performance/performance.cpp
+    renderer/command/performance/performance.h
+    renderer/command/resample/downmix_6ch_to_2ch.cpp
+    renderer/command/resample/downmix_6ch_to_2ch.h
+    renderer/command/resample/resample.h
+    renderer/command/resample/resample.cpp
+    renderer/command/resample/upsample.cpp
+    renderer/command/resample/upsample.h
+    renderer/command/sink/device.cpp
+    renderer/command/sink/device.h
+    renderer/command/sink/circular_buffer.cpp
+    renderer/command/sink/circular_buffer.h
+    renderer/command/command_buffer.cpp
+    renderer/command/command_buffer.h
+    renderer/command/command_generator.cpp
+    renderer/command/command_generator.h
+    renderer/command/command_list_header.h
+    renderer/command/command_processing_time_estimator.cpp
+    renderer/command/command_processing_time_estimator.h
+    renderer/command/commands.h
+    renderer/command/icommand.h
+    renderer/effect/effect_aux_info.cpp
+    renderer/effect/effect_aux_info.h
+    renderer/effect/effect_biquad_filter_info.cpp
+    renderer/effect/effect_biquad_filter_info.h
+    renderer/effect/effect_buffer_mixer_info.cpp
+    renderer/effect/effect_buffer_mixer_info.h
+    renderer/effect/effect_capture_info.cpp
+    renderer/effect/effect_capture_info.h
+    renderer/effect/effect_context.cpp
+    renderer/effect/effect_context.h
+    renderer/effect/effect_delay_info.cpp
+    renderer/effect/effect_delay_info.h
+    renderer/effect/effect_i3dl2_info.cpp
+    renderer/effect/effect_i3dl2_info.h
+    renderer/effect/effect_reset.h
+    renderer/effect/effect_info_base.h
+    renderer/effect/effect_light_limiter_info.cpp
+    renderer/effect/effect_light_limiter_info.h
+    renderer/effect/effect_result_state.h
+    renderer/effect/effect_reverb_info.h
+    renderer/effect/effect_reverb_info.cpp
+    renderer/mix/mix_context.cpp
+    renderer/mix/mix_context.h
+    renderer/mix/mix_info.cpp
+    renderer/mix/mix_info.h
+    renderer/memory/address_info.h
+    renderer/memory/memory_pool_info.cpp
+    renderer/memory/memory_pool_info.h
+    renderer/memory/pool_mapper.cpp
+    renderer/memory/pool_mapper.h
+    renderer/nodes/bit_array.h
+    renderer/nodes/edge_matrix.cpp
+    renderer/nodes/edge_matrix.h
+    renderer/nodes/node_states.cpp
+    renderer/nodes/node_states.h
+    renderer/performance/detail_aspect.cpp
+    renderer/performance/detail_aspect.h
+    renderer/performance/entry_aspect.cpp
+    renderer/performance/entry_aspect.h
+    renderer/performance/performance_detail.h
+    renderer/performance/performance_entry.h
+    renderer/performance/performance_entry_addresses.h
+    renderer/performance/performance_frame_header.h
+    renderer/performance/performance_manager.cpp
+    renderer/performance/performance_manager.h
+    renderer/sink/circular_buffer_sink_info.cpp
+    renderer/sink/circular_buffer_sink_info.h
+    renderer/sink/device_sink_info.cpp
+    renderer/sink/device_sink_info.h
+    renderer/sink/sink_context.cpp
+    renderer/sink/sink_context.h
+    renderer/sink/sink_info_base.cpp
+    renderer/sink/sink_info_base.h
+    renderer/splitter/splitter_context.cpp
+    renderer/splitter/splitter_context.h
+    renderer/splitter/splitter_destinations_data.cpp
+    renderer/splitter/splitter_destinations_data.h
+    renderer/splitter/splitter_info.cpp
+    renderer/splitter/splitter_info.h
+    renderer/system.cpp
+    renderer/system.h
+    renderer/system_manager.cpp
+    renderer/system_manager.h
+    renderer/upsampler/upsampler_info.h
+    renderer/upsampler/upsampler_manager.cpp
+    renderer/upsampler/upsampler_manager.h
+    renderer/upsampler/upsampler_state.h
+    renderer/voice/voice_channel_resource.h
+    renderer/voice/voice_context.cpp
+    renderer/voice/voice_context.h
+    renderer/voice/voice_info.cpp
+    renderer/voice/voice_info.h
+    renderer/voice/voice_state.h
+    sink/cubeb_sink.cpp
+    sink/cubeb_sink.h
+    sink/null_sink.h
+    sink/sdl2_sink.cpp
+    sink/sdl2_sink.h
+    sink/sink.h
+    sink/sink_details.cpp
+    sink/sink_details.h
+    sink/sink_stream.h
 )
 
 create_target_directory_groups(audio_core)
 
-if (NOT MSVC)
+if (MSVC)
+    target_compile_options(audio_core PRIVATE
+        /we4242 # 'identifier': conversion from 'type1' to 'type2', possible loss of data
+        /we4244 # 'conversion': conversion from 'type1' to 'type2', possible loss of data
+        /we4245 # 'conversion': conversion from 'type1' to 'type2', signed/unsigned mismatch
+        /we4254 # 'operator': conversion from 'type1:field_bits' to 'type2:field_bits', possible loss of data
+        /we4456 # Declaration of 'identifier' hides previous local declaration
+        /we4457 # Declaration of 'identifier' hides function parameter
+        /we4458 # Declaration of 'identifier' hides class member
+        /we4459 # Declaration of 'identifier' hides global declaration
+    )
+else()
     target_compile_options(audio_core PRIVATE
         -Werror=conversion
         -Werror=ignored-qualifiers
+        -Werror=shadow
+        -Werror=unused-variable
 
         $<$<CXX_COMPILER_ID:GNU>:-Werror=unused-but-set-parameter>
         $<$<CXX_COMPILER_ID:GNU>:-Werror=unused-but-set-variable>
@@ -58,6 +218,9 @@ if (NOT MSVC)
 endif()
 
 target_link_libraries(audio_core PUBLIC common core)
+if (ARCHITECTURE_x86_64)
+    target_link_libraries(audio_core PRIVATE dynarmic)
+endif()
 
 if(ENABLE_CUBEB)
     target_link_libraries(audio_core PRIVATE cubeb)

src/audio_core/audio_core.cpp

@@ -0,0 +1,60 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_core.h"
+#include "audio_core/sink/sink_details.h"
+#include "common/settings.h"
+#include "core/core.h"
+
+namespace AudioCore {
+
+AudioCore::AudioCore(Core::System& system) : audio_manager{std::make_unique<AudioManager>(system)} {
+    CreateSinks();
+    // Must be created after the sinks
+    adsp = std::make_unique<AudioRenderer::ADSP::ADSP>(system, *output_sink);
+}
+
+AudioCore ::~AudioCore() {
+    Shutdown();
+}
+
+void AudioCore::CreateSinks() {
+    const auto& sink_id{Settings::values.sink_id};
+    const auto& audio_output_device_id{Settings::values.audio_output_device_id};
+    const auto& audio_input_device_id{Settings::values.audio_input_device_id};
+
+    output_sink = Sink::CreateSinkFromID(sink_id.GetValue(), audio_output_device_id.GetValue());
+    input_sink = Sink::CreateSinkFromID(sink_id.GetValue(), audio_input_device_id.GetValue());
+}
+
+void AudioCore::Shutdown() {
+    audio_manager->Shutdown();
+}
+
+AudioManager& AudioCore::GetAudioManager() {
+    return *audio_manager;
+}
+
+Sink::Sink& AudioCore::GetOutputSink() {
+    return *output_sink;
+}
+
+Sink::Sink& AudioCore::GetInputSink() {
+    return *input_sink;
+}
+
+AudioRenderer::ADSP::ADSP& AudioCore::GetADSP() {
+    return *adsp;
+}
+
+void AudioCore::PauseSinks(const bool pausing) const {
+    if (pausing) {
+        output_sink->PauseStreams();
+        input_sink->PauseStreams();
+    } else {
+        output_sink->UnpauseStreams();
+        input_sink->UnpauseStreams();
+    }
+}
+
+} // namespace AudioCore

src/audio_core/audio_core.h

@@ -0,0 +1,84 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <memory>
+
+#include "audio_core/audio_manager.h"
+#include "audio_core/renderer/adsp/adsp.h"
+#include "audio_core/sink/sink.h"
+
+namespace Core {
+class System;
+}
+
+namespace AudioCore {
+
+class AudioManager;
+/**
+ * Main audio class, sotred inside the core, and holding the audio manager, all sinks, and the ADSP.
+ */
+class AudioCore {
+public:
+    explicit AudioCore(Core::System& system);
+    ~AudioCore();
+
+    /**
+     * Shutdown the audio core.
+     */
+    void Shutdown();
+
+    /**
+     * Get a reference to the audio manager.
+     *
+     * @return Ref to the audio manager.
+     */
+    AudioManager& GetAudioManager();
+
+    /**
+     * Get the audio output sink currently in use.
+     *
+     * @return Ref to the sink.
+     */
+    Sink::Sink& GetOutputSink();
+
+    /**
+     * Get the audio input sink currently in use.
+     *
+     * @return Ref to the sink.
+     */
+    Sink::Sink& GetInputSink();
+
+    /**
+     * Get the ADSP.
+     *
+     * @return Ref to the ADSP.
+     */
+    AudioRenderer::ADSP::ADSP& GetADSP();
+
+    /**
+     * Pause the sink. Called from the core.
+     *
+     * @param pausing - Is this pause due to an actual pause, or shutdown?
+     *                  Unfortunately, shutdown also pauses streams, which can cause issues.
+     */
+    void PauseSinks(bool pausing) const;
+
+private:
+    /**
+     * Create the sinks on startup.
+     */
+    void CreateSinks();
+
+    /// Main audio manager for audio in/out
+    std::unique_ptr<AudioManager> audio_manager;
+    /// Sink used for audio renderer and audio out
+    std::unique_ptr<Sink::Sink> output_sink;
+    /// Sink used for audio input
+    std::unique_ptr<Sink::Sink> input_sink;
+    /// The ADSP in the sysmodule
+    std::unique_ptr<AudioRenderer::ADSP::ADSP> adsp;
+};
+
+} // namespace AudioCore

src/audio_core/audio_event.cpp

@@ -0,0 +1,61 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_event.h"
+#include "common/assert.h"
+
+namespace AudioCore {
+
+size_t Event::GetManagerIndex(const Type type) const {
+    switch (type) {
+    case Type::AudioInManager:
+        return 0;
+    case Type::AudioOutManager:
+        return 1;
+    case Type::FinalOutputRecorderManager:
+        return 2;
+    case Type::Max:
+        return 3;
+    default:
+        UNREACHABLE();
+    }
+    return 3;
+}
+
+void Event::SetAudioEvent(const Type type, const bool signalled) {
+    events_signalled[GetManagerIndex(type)] = signalled;
+    if (signalled) {
+        manager_event.notify_one();
+    }
+}
+
+bool Event::CheckAudioEventSet(const Type type) const {
+    return events_signalled[GetManagerIndex(type)];
+}
+
+std::mutex& Event::GetAudioEventLock() {
+    return event_lock;
+}
+
+std::condition_variable_any& Event::GetAudioEvent() {
+    return manager_event;
+}
+
+bool Event::Wait(std::unique_lock<std::mutex>& l, const std::chrono::seconds timeout) {
+    bool timed_out{false};
+    if (!manager_event.wait_for(l, timeout, [&]() {
+            return std::ranges::any_of(events_signalled, [](bool x) { return x; });
+        })) {
+        timed_out = true;
+    }
+    return timed_out;
+}
+
+void Event::ClearEvents() {
+    events_signalled[0] = false;
+    events_signalled[1] = false;
+    events_signalled[2] = false;
+    events_signalled[3] = false;
+}
+
+} // namespace AudioCore

src/audio_core/audio_event.h

@@ -0,0 +1,92 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <atomic>
+#include <chrono>
+#include <condition_variable>
+#include <mutex>
+
+namespace AudioCore {
+/**
+ * Responsible for the input/output events, set by the stream backend when buffers are consumed, and
+ * waited on by the audio manager. These callbacks signal the game's events to keep the audio buffer
+ * recycling going.
+ * In a real Switch this is not a seprate class, and exists entirely within the audio manager.
+ * On the Switch it's implemented more simply through a MultiWaitEventHolder, where it can
+ * wait on multiple events at once, and the events are not needed by the backend.
+ */
+class Event {
+public:
+    enum class Type {
+        AudioInManager,
+        AudioOutManager,
+        FinalOutputRecorderManager,
+        Max,
+    };
+
+    /**
+     * Convert a manager type to an index.
+     *
+     * @param type - The manager type to convert
+     * @return The index of the type.
+     */
+    size_t GetManagerIndex(Type type) const;
+
+    /**
+     * Set an audio event to true or false.
+     *
+     * @param type      - The manager type to signal.
+     * @param signalled - Its signal state.
+     */
+    void SetAudioEvent(Type type, bool signalled);
+
+    /**
+     * Check if the given manager type is signalled.
+     *
+     * @param type - The manager type to check.
+     * @return True if the event is signalled, otherwise false.
+     */
+    bool CheckAudioEventSet(Type type) const;
+
+    /**
+     * Get the lock for audio events.
+     *
+     * @return Reference to the lock.
+     */
+    std::mutex& GetAudioEventLock();
+
+    /**
+     * Get the manager event, this signals the audio manager to release buffers and signal the game
+     * for more.
+     *
+     * @return Reference to the condition variable.
+     */
+    std::condition_variable_any& GetAudioEvent();
+
+    /**
+     * Wait on the manager_event.
+     *
+     * @param l       - Lock held by the wait.
+     * @param timeout - Timeout for the wait. This is 2 seconds by default.
+     * @return True if the wait timed out, otherwise false if signalled.
+     */
+    bool Wait(std::unique_lock<std::mutex>& l, std::chrono::seconds timeout);
+
+    /**
+     * Reset all manager events.
+     */
+    void ClearEvents();
+
+private:
+    /// Lock, used bythe audio manager
+    std::mutex event_lock;
+    /// Array of events, one per system type (see Type), last event is used to terminate
+    std::array<std::atomic<bool>, 4> events_signalled;
+    /// Event to signal the audio manager
+    std::condition_variable_any manager_event;
+};
+
+} // namespace AudioCore

src/audio_core/audio_in_manager.cpp

@@ -0,0 +1,86 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_core.h"
+#include "audio_core/audio_in_manager.h"
+#include "audio_core/audio_manager.h"
+#include "audio_core/in/audio_in.h"
+#include "core/core.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace AudioCore::AudioIn {
+
+Manager::Manager(Core::System& system_) : system{system_} {
+    std::iota(session_ids.begin(), session_ids.end(), 0);
+    num_free_sessions = MaxInSessions;
+}
+
+Result Manager::AcquireSessionId(size_t& session_id) {
+    if (num_free_sessions == 0) {
+        LOG_ERROR(Service_Audio, "All 4 AudioIn sessions are in use, cannot create any more");
+        return Service::Audio::ERR_MAXIMUM_SESSIONS_REACHED;
+    }
+    session_id = session_ids[next_session_id];
+    next_session_id = (next_session_id + 1) % MaxInSessions;
+    num_free_sessions--;
+    return ResultSuccess;
+}
+
+void Manager::ReleaseSessionId(const size_t session_id) {
+    std::scoped_lock l{mutex};
+    LOG_DEBUG(Service_Audio, "Freeing AudioIn session {}", session_id);
+    session_ids[free_session_id] = session_id;
+    num_free_sessions++;
+    free_session_id = (free_session_id + 1) % MaxInSessions;
+    sessions[session_id].reset();
+    applet_resource_user_ids[session_id] = 0;
+}
+
+Result Manager::LinkToManager() {
+    std::scoped_lock l{mutex};
+    if (!linked_to_manager) {
+        AudioManager& manager{system.AudioCore().GetAudioManager()};
+        manager.SetInManager(std::bind(&Manager::BufferReleaseAndRegister, this));
+        linked_to_manager = true;
+    }
+
+    return ResultSuccess;
+}
+
+void Manager::Start() {
+    if (sessions_started) {
+        return;
+    }
+
+    std::scoped_lock l{mutex};
+    for (auto& session : sessions) {
+        if (session) {
+            session->StartSession();
+        }
+    }
+
+    sessions_started = true;
+}
+
+void Manager::BufferReleaseAndRegister() {
+    std::scoped_lock l{mutex};
+    for (auto& session : sessions) {
+        if (session != nullptr) {
+            session->ReleaseAndRegisterBuffers();
+        }
+    }
+}
+
+u32 Manager::GetDeviceNames(std::vector<AudioRenderer::AudioDevice::AudioDeviceName>& names,
+                            [[maybe_unused]] const u32 max_count,
+                            [[maybe_unused]] const bool filter) {
+    std::scoped_lock l{mutex};
+
+    LinkToManager();
+
+    names.push_back(AudioRenderer::AudioDevice::AudioDeviceName("Uac"));
+
+    return 1;
+}
+
+} // namespace AudioCore::AudioIn

src/audio_core/audio_in_manager.h

@@ -0,0 +1,92 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <mutex>
+#include <vector>
+
+#include "audio_core/renderer/audio_device.h"
+
+namespace Core {
+class System;
+}
+
+namespace AudioCore::AudioIn {
+class In;
+
+constexpr size_t MaxInSessions = 4;
+/**
+ * Manages all audio in sessions.
+ */
+class Manager {
+public:
+    explicit Manager(Core::System& system);
+
+    /**
+     * Acquire a free session id for opening a new audio in.
+     *
+     * @param session_id - Output session_id.
+     * @return Result code.
+     */
+    Result AcquireSessionId(size_t& session_id);
+
+    /**
+     * Release a session id on close.
+     *
+     * @param session_id - Session id to free.
+     */
+    void ReleaseSessionId(size_t session_id);
+
+    /**
+     * Link the audio in manager to the main audio manager.
+     *
+     * @return Result code.
+     */
+    Result LinkToManager();
+
+    /**
+     * Start the audio in manager.
+     */
+    void Start();
+
+    /**
+     * Callback function, called by the audio manager when the audio in event is signalled.
+     */
+    void BufferReleaseAndRegister();
+
+    /**
+     * Get a list of audio in device names.
+     *
+     * @oaram names     - Output container to write names to.
+     * @param max_count - Maximum numebr of deivce names to write. Unused
+     * @param filter    - Should the list be filtered? Unused.
+     * @return Number of names written.
+     */
+    u32 GetDeviceNames(std::vector<AudioRenderer::AudioDevice::AudioDeviceName>& names,
+                       u32 max_count, bool filter);
+
+    /// Core system
+    Core::System& system;
+    /// Array of session ids
+    std::array<size_t, MaxInSessions> session_ids{};
+    /// Array of resource user ids
+    std::array<size_t, MaxInSessions> applet_resource_user_ids{};
+    /// Pointer to each open session
+    std::array<std::shared_ptr<In>, MaxInSessions> sessions{};
+    /// The number of free sessions
+    size_t num_free_sessions{};
+    /// The next session id to be taken
+    size_t next_session_id{};
+    /// The next session id to be freed
+    size_t free_session_id{};
+    /// Whether this is linked to the audio manager
+    bool linked_to_manager{};
+    /// Whether the sessions have been started
+    bool sessions_started{};
+    /// Protect state due to audio manager callback
+    std::recursive_mutex mutex{};
+};
+
+} // namespace AudioCore::AudioIn

src/audio_core/audio_manager.cpp

@@ -0,0 +1,80 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_in_manager.h"
+#include "audio_core/audio_manager.h"
+#include "audio_core/audio_out_manager.h"
+#include "core/core.h"
+
+namespace AudioCore {
+
+AudioManager::AudioManager(Core::System& system_) : system{system_} {
+    thread = std::jthread([this]() { ThreadFunc(); });
+}
+
+void AudioManager::Shutdown() {
+    running = false;
+    events.SetAudioEvent(Event::Type::Max, true);
+    thread.join();
+}
+
+Result AudioManager::SetOutManager(BufferEventFunc buffer_func) {
+    if (!running) {
+        return Service::Audio::ERR_OPERATION_FAILED;
+    }
+
+    std::scoped_lock l{lock};
+
+    const auto index{events.GetManagerIndex(Event::Type::AudioOutManager)};
+    if (buffer_events[index] == nullptr) {
+        buffer_events[index] = buffer_func;
+        needs_update = true;
+        events.SetAudioEvent(Event::Type::AudioOutManager, true);
+    }
+    return ResultSuccess;
+}
+
+Result AudioManager::SetInManager(BufferEventFunc buffer_func) {
+    if (!running) {
+        return Service::Audio::ERR_OPERATION_FAILED;
+    }
+
+    std::scoped_lock l{lock};
+
+    const auto index{events.GetManagerIndex(Event::Type::AudioInManager)};
+    if (buffer_events[index] == nullptr) {
+        buffer_events[index] = buffer_func;
+        needs_update = true;
+        events.SetAudioEvent(Event::Type::AudioInManager, true);
+    }
+    return ResultSuccess;
+}
+
+void AudioManager::SetEvent(const Event::Type type, const bool signalled) {
+    events.SetAudioEvent(type, signalled);
+}
+
+void AudioManager::ThreadFunc() {
+    std::unique_lock l{events.GetAudioEventLock()};
+    events.ClearEvents();
+    running = true;
+
+    while (running) {
+        auto timed_out{events.Wait(l, std::chrono::seconds(2))};
+
+        if (events.CheckAudioEventSet(Event::Type::Max)) {
+            break;
+        }
+
+        for (size_t i = 0; i < buffer_events.size(); i++) {
+            if (events.CheckAudioEventSet(Event::Type(i)) || timed_out) {
+                if (buffer_events[i]) {
+                    buffer_events[i]();
+                }
+            }
+            events.SetAudioEvent(Event::Type(i), false);
+        }
+    }
+}
+
+} // namespace AudioCore

src/audio_core/audio_manager.h

@@ -0,0 +1,101 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <atomic>
+#include <functional>
+#include <mutex>
+#include <thread>
+
+#include "audio_core/audio_event.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace Core {
+class System;
+}
+
+namespace AudioCore {
+
+namespace AudioOut {
+class Manager;
+}
+
+namespace AudioIn {
+class Manager;
+}
+
+/**
+ * The AudioManager's main purpose is to wait for buffer events for the audio in and out managers,
+ * and call an associated callback to release buffers.
+ *
+ * Execution pattern is:
+ *     Buffers appended ->
+ *     Buffers queued and played by the backend stream ->
+ *     When consumed, set the corresponding manager event and signal the audio manager ->
+ *     Consumed buffers are released, game is signalled ->
+ *     Game appends more buffers.
+ *
+ * This is only used by audio in and audio out.
+ */
+class AudioManager {
+    using BufferEventFunc = std::function<void()>;
+
+public:
+    explicit AudioManager(Core::System& system);
+
+    /**
+     * Shutdown the audio manager.
+     */
+    void Shutdown();
+
+    /**
+     * Register the out manager, keeping a function to be called when the out event is signalled.
+     *
+     * @param buffer_func - Function to be called on signal.
+     * @return Result code.
+     */
+    Result SetOutManager(BufferEventFunc buffer_func);
+
+    /**
+     * Register the in manager, keeping a function to be called when the in event is signalled.
+     *
+     * @param buffer_func - Function to be called on signal.
+     * @return Result code.
+     */
+    Result SetInManager(BufferEventFunc buffer_func);
+
+    /**
+     * Set an event to signalled, and signal the thread.
+     *
+     * @param type      - Manager type to set.
+     * @param signalled - Set the event to true or false?
+     */
+    void SetEvent(Event::Type type, bool signalled);
+
+private:
+    /**
+     * Main thread, waiting on a manager signal and calling the registered fucntion.
+     */
+    void ThreadFunc();
+
+    /// Core system
+    Core::System& system;
+    /// Have sessions started palying?
+    bool sessions_started{};
+    /// Is the main thread running?
+    std::atomic<bool> running{};
+    /// Unused
+    bool needs_update{};
+    /// Events to be set and signalled
+    Event events{};
+    /// Callbacks for each manager
+    std::array<BufferEventFunc, 3> buffer_events{};
+    /// General lock
+    std::mutex lock{};
+    /// Main thread for waiting and callbacks
+    std::jthread thread;
+};
+
+} // namespace AudioCore

src/audio_core/audio_out_manager.cpp

@@ -0,0 +1,81 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_core.h"
+#include "audio_core/audio_manager.h"
+#include "audio_core/audio_out_manager.h"
+#include "audio_core/out/audio_out.h"
+#include "core/core.h"
+#include "core/hle/kernel/k_event.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace AudioCore::AudioOut {
+
+Manager::Manager(Core::System& system_) : system{system_} {
+    std::iota(session_ids.begin(), session_ids.end(), 0);
+    num_free_sessions = MaxOutSessions;
+}
+
+Result Manager::AcquireSessionId(size_t& session_id) {
+    if (num_free_sessions == 0) {
+        LOG_ERROR(Service_Audio, "All 12 Audio Out sessions are in use, cannot create any more");
+        return Service::Audio::ERR_MAXIMUM_SESSIONS_REACHED;
+    }
+    session_id = session_ids[next_session_id];
+    next_session_id = (next_session_id + 1) % MaxOutSessions;
+    num_free_sessions--;
+    return ResultSuccess;
+}
+
+void Manager::ReleaseSessionId(const size_t session_id) {
+    std::scoped_lock l{mutex};
+    LOG_DEBUG(Service_Audio, "Freeing AudioOut session {}", session_id);
+    session_ids[free_session_id] = session_id;
+    num_free_sessions++;
+    free_session_id = (free_session_id + 1) % MaxOutSessions;
+    sessions[session_id].reset();
+    applet_resource_user_ids[session_id] = 0;
+}
+
+Result Manager::LinkToManager() {
+    std::scoped_lock l{mutex};
+    if (!linked_to_manager) {
+        AudioManager& manager{system.AudioCore().GetAudioManager()};
+        manager.SetOutManager(std::bind(&Manager::BufferReleaseAndRegister, this));
+        linked_to_manager = true;
+    }
+
+    return ResultSuccess;
+}
+
+void Manager::Start() {
+    if (sessions_started) {
+        return;
+    }
+
+    std::scoped_lock l{mutex};
+    for (auto& session : sessions) {
+        if (session) {
+            session->StartSession();
+        }
+    }
+
+    sessions_started = true;
+}
+
+void Manager::BufferReleaseAndRegister() {
+    std::scoped_lock l{mutex};
+    for (auto& session : sessions) {
+        if (session != nullptr) {
+            session->ReleaseAndRegisterBuffers();
+        }
+    }
+}
+
+u32 Manager::GetAudioOutDeviceNames(
+    std::vector<AudioRenderer::AudioDevice::AudioDeviceName>& names) const {
+    names.push_back({"DeviceOut"});
+    return 1;
+}
+
+} // namespace AudioCore::AudioOut

src/audio_core/audio_out_manager.h

@@ -0,0 +1,89 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <mutex>
+
+#include "audio_core/renderer/audio_device.h"
+
+namespace Core {
+class System;
+}
+
+namespace AudioCore::AudioOut {
+class Out;
+
+constexpr size_t MaxOutSessions = 12;
+/**
+ * Manages all audio out sessions.
+ */
+class Manager {
+public:
+    explicit Manager(Core::System& system);
+
+    /**
+     * Acquire a free session id for opening a new audio out.
+     *
+     * @param session_id - Output session_id.
+     * @return Result code.
+     */
+    Result AcquireSessionId(size_t& session_id);
+
+    /**
+     * Release a session id on close.
+     *
+     * @param session_id - Session id to free.
+     */
+    void ReleaseSessionId(size_t session_id);
+
+    /**
+     * Link this manager to the main audio manager.
+     *
+     * @return Result code.
+     */
+    Result LinkToManager();
+
+    /**
+     * Start the audio out manager.
+     */
+    void Start();
+
+    /**
+     * Callback function, called by the audio manager when the audio out event is signalled.
+     */
+    void BufferReleaseAndRegister();
+
+    /**
+     * Get a list of audio out device names.
+     *
+     * @oaram names     - Output container to write names to.
+     * @return Number of names written.
+     */
+    u32 GetAudioOutDeviceNames(
+        std::vector<AudioRenderer::AudioDevice::AudioDeviceName>& names) const;
+
+    /// Core system
+    Core::System& system;
+    /// Array of session ids
+    std::array<size_t, MaxOutSessions> session_ids{};
+    /// Array of resource user ids
+    std::array<size_t, MaxOutSessions> applet_resource_user_ids{};
+    /// Pointer to each open session
+    std::array<std::shared_ptr<Out>, MaxOutSessions> sessions{};
+    /// The number of free sessions
+    size_t num_free_sessions{};
+    /// The next session id to be taken
+    size_t next_session_id{};
+    /// The next session id to be freed
+    size_t free_session_id{};
+    /// Whether this is linked to the audio manager
+    bool linked_to_manager{};
+    /// Whether the sessions have been started
+    bool sessions_started{};
+    /// Protect state due to audio manager callback
+    std::recursive_mutex mutex{};
+};
+
+} // namespace AudioCore::AudioOut

src/audio_core/audio_render_manager.cpp

@@ -0,0 +1,70 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_render_manager.h"
+#include "audio_core/common/audio_renderer_parameter.h"
+#include "audio_core/common/feature_support.h"
+#include "core/core.h"
+
+namespace AudioCore::AudioRenderer {
+
+Manager::Manager(Core::System& system_)
+    : system{system_}, system_manager{std::make_unique<SystemManager>(system)} {
+    std::iota(session_ids.begin(), session_ids.end(), 0);
+}
+
+Manager::~Manager() {
+    Stop();
+}
+
+void Manager::Stop() {
+    system_manager->Stop();
+}
+
+SystemManager& Manager::GetSystemManager() {
+    return *system_manager;
+}
+
+auto Manager::GetWorkBufferSize(const AudioRendererParameterInternal& params, u64& out_count)
+    -> Result {
+    if (!CheckValidRevision(params.revision)) {
+        return Service::Audio::ERR_INVALID_REVISION;
+    }
+
+    out_count = System::GetWorkBufferSize(params);
+
+    return ResultSuccess;
+}
+
+s32 Manager::GetSessionId() {
+    std::scoped_lock l{session_lock};
+    auto session_id{session_ids[session_count]};
+
+    if (session_id == -1) {
+        return -1;
+    }
+
+    session_ids[session_count] = -1;
+    session_count++;
+    return session_id;
+}
+
+void Manager::ReleaseSessionId(const s32 session_id) {
+    std::scoped_lock l{session_lock};
+    session_ids[--session_count] = session_id;
+}
+
+u32 Manager::GetSessionCount() {
+    std::scoped_lock l{session_lock};
+    return session_count;
+}
+
+bool Manager::AddSystem(System& system_) {
+    return system_manager->Add(system_);
+}
+
+bool Manager::RemoveSystem(System& system_) {
+    return system_manager->Remove(system_);
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/audio_render_manager.h

@@ -0,0 +1,103 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <memory>
+#include <mutex>
+
+#include "audio_core/common/common.h"
+#include "audio_core/renderer/system_manager.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace Core {
+class System;
+}
+
+namespace AudioCore {
+struct AudioRendererParameterInternal;
+
+namespace AudioRenderer {
+/**
+ * Wrapper for the audio system manager, handles service calls.
+ */
+class Manager {
+public:
+    explicit Manager(Core::System& system);
+    ~Manager();
+
+    /**
+     * Stop the manager.
+     */
+    void Stop();
+
+    /**
+     * Get the system manager.
+     *
+     * @return The system manager.
+     */
+    SystemManager& GetSystemManager();
+
+    /**
+     * Get required size for the audio renderer workbuffer.
+     *
+     * @param params    - Input parameters with the numbers of voices/mixes/sinks etc.
+     * @param out_count - Output size of the required workbuffer.
+     * @return Result code.
+     */
+    Result GetWorkBufferSize(const AudioRendererParameterInternal& params, u64& out_count);
+
+    /**
+     * Get a new session id.
+     *
+     * @return The new session id. -1 if invalid, otherwise 0-MaxRendererSessions.
+     */
+    s32 GetSessionId();
+
+    /**
+     * Get the number of currently active sessions.
+     *
+     * @return The number of active sessions.
+     */
+    u32 GetSessionCount();
+
+    /**
+     * Add a renderer system to the manager.
+     * The system will be reguarly called to generate commands for the AudioRenderer.
+     *
+     * @param system - The system to add.
+     * @return True if the system was sucessfully added, otherwise false.
+     */
+    bool AddSystem(System& system);
+
+    /**
+     * Remove a renderer system from the manager.
+     *
+     * @param system - The system to remove.
+     * @return True if the system was sucessfully removed, otherwise false.
+     */
+    bool RemoveSystem(System& system);
+
+    /**
+     * Free a session id when the system wants to shut down.
+     *
+     * @param session_id - The session id to free.
+     */
+    void ReleaseSessionId(s32 session_id);
+
+private:
+    /// Core system
+    Core::System& system;
+    /// Session ids, -1 when in use
+    std::array<s32, MaxRendererSessions> session_ids{};
+    /// Number of active renderers
+    u32 session_count{};
+    /// Lock for interacting with the sessions
+    std::mutex session_lock{};
+    /// Regularly generates commands from the registered systems for the AudioRenderer
+    std::unique_ptr<SystemManager> system_manager{};
+};
+
+} // namespace AudioRenderer
+} // namespace AudioCore

src/audio_core/common/audio_renderer_parameter.h

@@ -0,0 +1,60 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <span>
+
+#include "audio_core/renderer/behavior/behavior_info.h"
+#include "audio_core/renderer/memory/memory_pool_info.h"
+#include "audio_core/renderer/upsampler/upsampler_manager.h"
+#include "common/common_types.h"
+
+namespace AudioCore {
+/**
+ * Execution mode of the audio renderer.
+ * Only Auto is currently supported.
+ */
+enum class ExecutionMode : u8 {
+    Auto,
+    Manual,
+};
+
+/**
+ * Parameters from the game, passed to the audio renderer for initialisation.
+ */
+struct AudioRendererParameterInternal {
+    /* 0x00 */ u32 sample_rate;
+    /* 0x04 */ u32 sample_count;
+    /* 0x08 */ u32 mixes;
+    /* 0x0C */ u32 sub_mixes;
+    /* 0x10 */ u32 voices;
+    /* 0x14 */ u32 sinks;
+    /* 0x18 */ u32 effects;
+    /* 0x1C */ u32 perf_frames;
+    /* 0x20 */ u16 voice_drop_enabled;
+    /* 0x22 */ u8 rendering_device;
+    /* 0x23 */ ExecutionMode execution_mode;
+    /* 0x24 */ u32 splitter_infos;
+    /* 0x28 */ s32 splitter_destinations;
+    /* 0x2C */ u32 external_context_size;
+    /* 0x30 */ u32 revision;
+    /* 0x34 */ char unk34[0x4];
+};
+static_assert(sizeof(AudioRendererParameterInternal) == 0x38,
+              "AudioRendererParameterInternal has the wrong size!");
+
+/**
+ * Context for rendering, contains a bunch of useful fields for the command generator.
+ */
+struct AudioRendererSystemContext {
+    s32 session_id;
+    s8 channels;
+    s16 mix_buffer_count;
+    AudioRenderer::BehaviorInfo* behavior;
+    std::span<s32> depop_buffer;
+    AudioRenderer::UpsamplerManager* upsampler_manager;
+    AudioRenderer::MemoryPoolInfo* memory_pool_info;
+};
+
+} // namespace AudioCore

src/audio_core/common/common.h

@@ -0,0 +1,97 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <numeric>
+
+#include "common/assert.h"
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+
+namespace AudioCore {
+using CpuAddr = std::uintptr_t;
+
+enum class PlayState : u8 {
+    Started,
+    Stopped,
+    Paused,
+};
+
+enum class SrcQuality : u8 {
+    Medium,
+    High,
+    Low,
+};
+
+enum class SampleFormat : u8 {
+    Invalid,
+    PcmInt8,
+    PcmInt16,
+    PcmInt24,
+    PcmInt32,
+    PcmFloat,
+    Adpcm,
+};
+
+enum class SessionTypes {
+    AudioIn,
+    AudioOut,
+    FinalOutputRecorder,
+};
+
+constexpr u32 BufferCount = 32;
+
+constexpr u32 MaxRendererSessions = 2;
+constexpr u32 TargetSampleCount = 240;
+constexpr u32 TargetSampleRate = 48'000;
+constexpr u32 MaxChannels = 6;
+constexpr u32 MaxMixBuffers = 24;
+constexpr u32 MaxWaveBuffers = 4;
+constexpr s32 LowestVoicePriority = 0xFF;
+constexpr s32 HighestVoicePriority = 0;
+constexpr u32 BufferAlignment = 0x40;
+constexpr u32 WorkbufferAlignment = 0x1000;
+constexpr s32 FinalMixId = 0;
+constexpr s32 InvalidDistanceFromFinalMix = std::numeric_limits<s32>::min();
+constexpr s32 UnusedSplitterId = -1;
+constexpr s32 UnusedMixId = std::numeric_limits<s32>::max();
+constexpr u32 InvalidNodeId = 0xF0000000;
+constexpr s32 InvalidProcessOrder = -1;
+constexpr u32 MaxBiquadFilters = 2;
+constexpr u32 MaxEffects = 256;
+
+constexpr bool IsChannelCountValid(u16 channel_count) {
+    return channel_count <= 6 &&
+           (channel_count == 1 || channel_count == 2 || channel_count == 4 || channel_count == 6);
+}
+
+constexpr u32 GetSplitterInParamHeaderMagic() {
+    return Common::MakeMagic('S', 'N', 'D', 'H');
+}
+
+constexpr u32 GetSplitterInfoMagic() {
+    return Common::MakeMagic('S', 'N', 'D', 'I');
+}
+
+constexpr u32 GetSplitterSendDataMagic() {
+    return Common::MakeMagic('S', 'N', 'D', 'D');
+}
+
+constexpr size_t GetSampleFormatByteSize(SampleFormat format) {
+    switch (format) {
+    case SampleFormat::PcmInt8:
+        return 1;
+    case SampleFormat::PcmInt16:
+        return 2;
+    case SampleFormat::PcmInt24:
+        return 3;
+    case SampleFormat::PcmInt32:
+    case SampleFormat::PcmFloat:
+        return 4;
+    default:
+        return 2;
+    }
+}
+
+} // namespace AudioCore

src/audio_core/common/feature_support.h

@@ -0,0 +1,99 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <map>
+#include <ranges>
+#include <tuple>
+
+#include "common/assert.h"
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+
+namespace AudioCore {
+constexpr u32 CurrentRevision = 10;
+
+enum class SupportTags {
+    CommandProcessingTimeEstimatorVersion4,
+    CommandProcessingTimeEstimatorVersion3,
+    CommandProcessingTimeEstimatorVersion2,
+    MultiTapBiquadFilterProcessing,
+    EffectInfoVer2,
+    WaveBufferVer2,
+    BiquadFilterFloatProcessing,
+    VolumeMixParameterPrecisionQ23,
+    MixInParameterDirtyOnlyUpdate,
+    BiquadFilterEffectStateClearBugFix,
+    VoicePlayedSampleCountResetAtLoopPoint,
+    VoicePitchAndSrcSkipped,
+    SplitterBugFix,
+    FlushVoiceWaveBuffers,
+    ElapsedFrameCount,
+    AudioRendererVariadicCommandBufferSize,
+    PerformanceMetricsDataFormatVersion2,
+    AudioRendererProcessingTimeLimit80Percent,
+    AudioRendererProcessingTimeLimit75Percent,
+    AudioRendererProcessingTimeLimit70Percent,
+    AdpcmLoopContextBugFix,
+    Splitter,
+    LongSizePreDelay,
+    AudioUsbDeviceOutput,
+    DeviceApiVersion2,
+
+    // Not a real tag, just here to get the count.
+    Size
+};
+
+constexpr u32 GetRevisionNum(u32 user_revision) {
+    if (user_revision >= 0x100) {
+        user_revision -= Common::MakeMagic('R', 'E', 'V', '0');
+        user_revision >>= 24;
+    }
+    return user_revision;
+};
+
+constexpr bool CheckFeatureSupported(SupportTags tag, u32 user_revision) {
+    constexpr std::array<std::pair<SupportTags, u32>, static_cast<u32>(SupportTags::Size)> features{
+        {
+            {SupportTags::AudioRendererProcessingTimeLimit70Percent, 1},
+            {SupportTags::Splitter, 2},
+            {SupportTags::AdpcmLoopContextBugFix, 2},
+            {SupportTags::LongSizePreDelay, 3},
+            {SupportTags::AudioUsbDeviceOutput, 4},
+            {SupportTags::AudioRendererProcessingTimeLimit75Percent, 4},
+            {SupportTags::VoicePlayedSampleCountResetAtLoopPoint, 5},
+            {SupportTags::VoicePitchAndSrcSkipped, 5},
+            {SupportTags::SplitterBugFix, 5},
+            {SupportTags::FlushVoiceWaveBuffers, 5},
+            {SupportTags::ElapsedFrameCount, 5},
+            {SupportTags::AudioRendererProcessingTimeLimit80Percent, 5},
+            {SupportTags::AudioRendererVariadicCommandBufferSize, 5},
+            {SupportTags::PerformanceMetricsDataFormatVersion2, 5},
+            {SupportTags::CommandProcessingTimeEstimatorVersion2, 5},
+            {SupportTags::BiquadFilterEffectStateClearBugFix, 6},
+            {SupportTags::BiquadFilterFloatProcessing, 7},
+            {SupportTags::VolumeMixParameterPrecisionQ23, 7},
+            {SupportTags::MixInParameterDirtyOnlyUpdate, 7},
+            {SupportTags::WaveBufferVer2, 8},
+            {SupportTags::CommandProcessingTimeEstimatorVersion3, 8},
+            {SupportTags::EffectInfoVer2, 9},
+            {SupportTags::CommandProcessingTimeEstimatorVersion4, 10},
+            {SupportTags::MultiTapBiquadFilterProcessing, 10},
+        }};
+
+    const auto& feature =
+        std::ranges::find_if(features, [tag](const auto& entry) { return entry.first == tag; });
+    if (feature == features.cend()) {
+        LOG_ERROR(Service_Audio, "Invalid SupportTag {}!", static_cast<u32>(tag));
+        return false;
+    }
+    user_revision = GetRevisionNum(user_revision);
+    return (*feature).second <= user_revision;
+}
+
+constexpr bool CheckValidRevision(u32 user_revision) {
+    return GetRevisionNum(user_revision) <= CurrentRevision;
+};
+
+} // namespace AudioCore

src/audio_core/common/wave_buffer.h

@@ -0,0 +1,35 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace AudioCore {
+
+struct WaveBufferVersion1 {
+    CpuAddr buffer;
+    u64 buffer_size;
+    u32 start_offset;
+    u32 end_offset;
+    bool loop;
+    bool stream_ended;
+    CpuAddr context;
+    u64 context_size;
+};
+
+struct WaveBufferVersion2 {
+    CpuAddr buffer;
+    CpuAddr context;
+    u64 buffer_size;
+    u64 context_size;
+    u32 start_offset;
+    u32 end_offset;
+    u32 loop_start_offset;
+    u32 loop_end_offset;
+    s32 loop_count;
+    bool loop;
+    bool stream_ended;
+};
+
+} // namespace AudioCore

src/audio_core/common/workbuffer_allocator.h

@@ -0,0 +1,100 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <span>
+
+#include "common/alignment.h"
+#include "common/assert.h"
+#include "common/common_types.h"
+
+namespace AudioCore {
+/**
+ * Responsible for allocating up a workbuffer into multiple pieces.
+ * Takes in a buffer and size (it does not own them), and allocates up the buffer via Allocate.
+ */
+class WorkbufferAllocator {
+public:
+    explicit WorkbufferAllocator(std::span<u8> buffer_, u64 size_)
+        : buffer{reinterpret_cast<u64>(buffer_.data())}, size{size_} {}
+
+    /**
+     * Allocate the given count of T elements, aligned to alignment.
+     *
+     * @param count     - The number of elements to allocate.
+     * @param alignment - The required starting alignment.
+     * @return Non-owning container of allocated elements.
+     */
+    template <typename T>
+    std::span<T> Allocate(u64 count, u64 alignment) {
+        u64 out{0};
+        u64 byte_size{count * sizeof(T)};
+
+        if (byte_size > 0) {
+            auto current{buffer + offset};
+            auto aligned_buffer{Common::AlignUp(current, alignment)};
+            if (aligned_buffer + byte_size <= buffer + size) {
+                out = aligned_buffer;
+                offset = byte_size - buffer + aligned_buffer;
+            } else {
+                LOG_ERROR(
+                    Service_Audio,
+                    "Allocated buffer was too small to hold new alloc.\nAllocator size={:08X}, "
+                    "offset={:08X}.\nAttempting to allocate {:08X} with alignment={:02X}",
+                    size, offset, byte_size, alignment);
+                count = 0;
+            }
+        }
+
+        return std::span<T>(reinterpret_cast<T*>(out), count);
+    }
+
+    /**
+     * Align the current offset to the given alignment.
+     *
+     * @param alignment - The required starting alignment.
+     */
+    void Align(u64 alignment) {
+        auto current{buffer + offset};
+        auto aligned_buffer{Common::AlignUp(current, alignment)};
+        offset = 0 - buffer + aligned_buffer;
+    }
+
+    /**
+     * Get the current buffer offset.
+     *
+     * @return The current allocating offset.
+     */
+    u64 GetCurrentOffset() const {
+        return offset;
+    }
+
+    /**
+     * Get the current buffer size.
+     *
+     * @return The size of the current buffer.
+     */
+    u64 GetSize() const {
+        return size;
+    }
+
+    /**
+     * Get the remaining size that can be allocated.
+     *
+     * @return The remaining size left in the buffer.
+     */
+    u64 GetRemainingSize() const {
+        return size - offset;
+    }
+
+private:
+    /// The buffer into which we are allocating.
+    u64 buffer;
+    /// Size of the buffer we're allocating to.
+    u64 size;
+    /// Current offset into the buffer, an error will be thrown if it exceeds size.
+    u64 offset{};
+};
+
+} // namespace AudioCore

src/audio_core/device/audio_buffer.h

@@ -0,0 +1,21 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace AudioCore {
+
+struct AudioBuffer {
+    /// Timestamp this buffer completed playing.
+    s64 played_timestamp;
+    /// Game memory address for these samples.
+    VAddr samples;
+    /// Unqiue identifier for this buffer.
+    u64 tag;
+    /// Size of the samples buffer.
+    u64 size;
+};
+
+} // namespace AudioCore

src/audio_core/device/audio_buffers.h

@@ -0,0 +1,304 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <mutex>
+#include <span>
+#include <vector>
+
+#include "audio_buffer.h"
+#include "audio_core/device/device_session.h"
+#include "core/core_timing.h"
+
+namespace AudioCore {
+
+constexpr s32 BufferAppendLimit = 4;
+
+/**
+ * A ringbuffer of N audio buffers.
+ * The buffer contains 3 sections:
+ *     Appended   - Buffers added to the ring, but have yet to be sent to the audio backend.
+ *     Registered - Buffers sent to the backend and queued for playback.
+ *     Released   - Buffers which have been played, and can now be recycled.
+ * Any others are free/untracked.
+ *
+ * @tparam N - Maximum number of buffers in the ring.
+ */
+template <size_t N>
+class AudioBuffers {
+public:
+    explicit AudioBuffers(size_t limit) : append_limit{static_cast<u32>(limit)} {}
+
+    /**
+     * Append a new audio buffer to the ring.
+     *
+     * @param buffer - The new buffer.
+     */
+    void AppendBuffer(AudioBuffer& buffer) {
+        std::scoped_lock l{lock};
+        buffers[appended_index] = buffer;
+        appended_count++;
+        appended_index = (appended_index + 1) % append_limit;
+    }
+
+    /**
+     * Register waiting buffers, up to a maximum of BufferAppendLimit.
+     *
+     * @param out_buffers - The buffers which were registered.
+     */
+    void RegisterBuffers(std::vector<AudioBuffer>& out_buffers) {
+        std::scoped_lock l{lock};
+        const s32 to_register{std::min(std::min(appended_count, BufferAppendLimit),
+                                       BufferAppendLimit - registered_count)};
+
+        for (s32 i = 0; i < to_register; i++) {
+            s32 index{appended_index - appended_count};
+            if (index < 0) {
+                index += N;
+            }
+            out_buffers.push_back(buffers[index]);
+            registered_count++;
+            registered_index = (registered_index + 1) % append_limit;
+
+            appended_count--;
+            if (appended_count == 0) {
+                break;
+            }
+        }
+    }
+
+    /**
+     * Release a single buffer. Must be already registered.
+     *
+     * @param index     - The buffer index to release.
+     * @param timestamp - The released timestamp for this buffer.
+     */
+    void ReleaseBuffer(s32 index, s64 timestamp) {
+        std::scoped_lock l{lock};
+        buffers[index].played_timestamp = timestamp;
+
+        registered_count--;
+        released_count++;
+        released_index = (released_index + 1) % append_limit;
+    }
+
+    /**
+     * Release all registered buffers.
+     *
+     * @param timestamp - The released timestamp for this buffer.
+     * @return Is the buffer was released.
+     */
+    bool ReleaseBuffers(Core::Timing::CoreTiming& core_timing, DeviceSession& session) {
+        std::scoped_lock l{lock};
+        bool buffer_released{false};
+        while (registered_count > 0) {
+            auto index{registered_index - registered_count};
+            if (index < 0) {
+                index += N;
+            }
+
+            // Check with the backend if this buffer can be released yet.
+            if (!session.IsBufferConsumed(buffers[index].tag)) {
+                break;
+            }
+
+            ReleaseBuffer(index, core_timing.GetGlobalTimeNs().count());
+            buffer_released = true;
+        }
+
+        return buffer_released || registered_count == 0;
+    }
+
+    /**
+     * Get all released buffers.
+     *
+     * @param tags - Container to be filled with the released buffers' tags.
+     * @return The number of buffers released.
+     */
+    u32 GetReleasedBuffers(std::span<u64> tags) {
+        std::scoped_lock l{lock};
+        u32 released{0};
+
+        while (released_count > 0) {
+            auto index{released_index - released_count};
+            if (index < 0) {
+                index += N;
+            }
+
+            auto& buffer{buffers[index]};
+            released_count--;
+
+            auto tag{buffer.tag};
+            buffer.played_timestamp = 0;
+            buffer.samples = 0;
+            buffer.tag = 0;
+            buffer.size = 0;
+
+            if (tag == 0) {
+                break;
+            }
+
+            tags[released++] = tag;
+
+            if (released >= tags.size()) {
+                break;
+            }
+        }
+
+        return released;
+    }
+
+    /**
+     * Get all appended and registered buffers.
+     *
+     * @param buffers_flushed - Output vector for the buffers which are released.
+     * @param max_buffers     - Maximum number of buffers to released.
+     * @return The number of buffers released.
+     */
+    u32 GetRegisteredAppendedBuffers(std::vector<AudioBuffer>& buffers_flushed, u32 max_buffers) {
+        std::scoped_lock l{lock};
+        if (registered_count + appended_count == 0) {
+            return 0;
+        }
+
+        size_t buffers_to_flush{
+            std::min(static_cast<u32>(registered_count + appended_count), max_buffers)};
+        if (buffers_to_flush == 0) {
+            return 0;
+        }
+
+        while (registered_count > 0) {
+            auto index{registered_index - registered_count};
+            if (index < 0) {
+                index += N;
+            }
+
+            buffers_flushed.push_back(buffers[index]);
+
+            registered_count--;
+            released_count++;
+            released_index = (released_index + 1) % append_limit;
+
+            if (buffers_flushed.size() >= buffers_to_flush) {
+                break;
+            }
+        }
+
+        while (appended_count > 0) {
+            auto index{appended_index - appended_count};
+            if (index < 0) {
+                index += N;
+            }
+
+            buffers_flushed.push_back(buffers[index]);
+
+            appended_count--;
+            released_count++;
+            released_index = (released_index + 1) % append_limit;
+
+            if (buffers_flushed.size() >= buffers_to_flush) {
+                break;
+            }
+        }
+
+        return static_cast<u32>(buffers_flushed.size());
+    }
+
+    /**
+     * Check if the given tag is in the buffers.
+     *
+     * @param tag - Unique tag of the buffer to search for.
+     * @return True if the buffer is still in the ring, otherwise false.
+     */
+    bool ContainsBuffer(const u64 tag) const {
+        std::scoped_lock l{lock};
+        const auto registered_buffers{appended_count + registered_count + released_count};
+
+        if (registered_buffers == 0) {
+            return false;
+        }
+
+        auto index{released_index - released_count};
+        if (index < 0) {
+            index += append_limit;
+        }
+
+        for (s32 i = 0; i < registered_buffers; i++) {
+            if (buffers[index].tag == tag) {
+                return true;
+            }
+            index = (index + 1) % append_limit;
+        }
+
+        return false;
+    }
+
+    /**
+     * Get the number of active buffers in the ring.
+     * That is, appended, registered and released buffers.
+     *
+     * @return Number of active buffers.
+     */
+    u32 GetAppendedRegisteredCount() const {
+        std::scoped_lock l{lock};
+        return appended_count + registered_count;
+    }
+
+    /**
+     * Get the total number of active buffers in the ring.
+     * That is, appended, registered and released buffers.
+     *
+     * @return Number of active buffers.
+     */
+    u32 GetTotalBufferCount() const {
+        std::scoped_lock l{lock};
+        return static_cast<u32>(appended_count + registered_count + released_count);
+    }
+
+    /**
+     * Flush all of the currently appended and registered buffers
+     *
+     * @param buffers_released - Output count for the number of buffers released.
+     * @return True if buffers were successfully flushed, otherwise false.
+     */
+    bool FlushBuffers(u32& buffers_released) {
+        std::scoped_lock l{lock};
+        std::vector<AudioBuffer> buffers_flushed{};
+
+        buffers_released = GetRegisteredAppendedBuffers(buffers_flushed, append_limit);
+
+        if (registered_count > 0) {
+            return false;
+        }
+
+        if (static_cast<u32>(released_count + appended_count) > append_limit) {
+            return false;
+        }
+
+        return true;
+    }
+
+private:
+    /// Buffer lock
+    mutable std::recursive_mutex lock{};
+    /// The audio buffers
+    std::array<AudioBuffer, N> buffers{};
+    /// Current released index
+    s32 released_index{};
+    /// Number of released buffers
+    s32 released_count{};
+    /// Current registered index
+    s32 registered_index{};
+    /// Number of registered buffers
+    s32 registered_count{};
+    /// Current appended index
+    s32 appended_index{};
+    /// Number of appended buffers
+    s32 appended_count{};
+    /// Maximum number of buffers (default 32)
+    u32 append_limit{};
+};
+
+} // namespace AudioCore

src/audio_core/device/device_session.cpp

@@ -0,0 +1,111 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_core.h"
+#include "audio_core/audio_manager.h"
+#include "audio_core/device/audio_buffer.h"
+#include "audio_core/device/device_session.h"
+#include "audio_core/sink/sink_stream.h"
+#include "core/core.h"
+#include "core/memory.h"
+
+namespace AudioCore {
+
+DeviceSession::DeviceSession(Core::System& system_) : system{system_} {}
+
+DeviceSession::~DeviceSession() {
+    Finalize();
+}
+
+Result DeviceSession::Initialize(std::string_view name_, SampleFormat sample_format_,
+                                 u16 channel_count_, size_t session_id_, u32 handle_,
+                                 u64 applet_resource_user_id_, Sink::StreamType type_) {
+    if (stream) {
+        Finalize();
+    }
+    name = fmt::format("{}-{}", name_, session_id_);
+    type = type_;
+    sample_format = sample_format_;
+    channel_count = channel_count_;
+    session_id = session_id_;
+    handle = handle_;
+    applet_resource_user_id = applet_resource_user_id_;
+
+    if (type == Sink::StreamType::In) {
+        sink = &system.AudioCore().GetInputSink();
+    } else {
+        sink = &system.AudioCore().GetOutputSink();
+    }
+    stream = sink->AcquireSinkStream(system, channel_count, name, type);
+    return ResultSuccess;
+}
+
+void DeviceSession::Finalize() {
+    Stop();
+    sink->CloseStream(stream);
+    stream = nullptr;
+}
+
+void DeviceSession::Start() {
+    stream->SetPlayedSampleCount(played_sample_count);
+    stream->Start();
+}
+
+void DeviceSession::Stop() {
+    if (stream) {
+        played_sample_count = stream->GetPlayedSampleCount();
+        stream->Stop();
+    }
+}
+
+void DeviceSession::AppendBuffers(std::span<AudioBuffer> buffers) const {
+    auto& memory{system.Memory()};
+
+    for (size_t i = 0; i < buffers.size(); i++) {
+        Sink::SinkBuffer new_buffer{
+            .frames = buffers[i].size / (channel_count * sizeof(s16)),
+            .frames_played = 0,
+            .tag = buffers[i].tag,
+            .consumed = false,
+        };
+
+        if (type == Sink::StreamType::In) {
+            std::vector<s16> samples{};
+            stream->AppendBuffer(new_buffer, samples);
+        } else {
+            std::vector<s16> samples(buffers[i].size / sizeof(s16));
+            memory.ReadBlockUnsafe(buffers[i].samples, samples.data(), buffers[i].size);
+            stream->AppendBuffer(new_buffer, samples);
+        }
+    }
+}
+
+void DeviceSession::ReleaseBuffer(AudioBuffer& buffer) const {
+    if (type == Sink::StreamType::In) {
+        auto& memory{system.Memory()};
+        auto samples{stream->ReleaseBuffer(buffer.size / sizeof(s16))};
+        memory.WriteBlockUnsafe(buffer.samples, samples.data(), buffer.size);
+    }
+}
+
+bool DeviceSession::IsBufferConsumed(u64 tag) const {
+    if (stream) {
+        return stream->IsBufferConsumed(tag);
+    }
+    return true;
+}
+
+void DeviceSession::SetVolume(f32 volume) const {
+    if (stream) {
+        stream->SetSystemVolume(volume);
+    }
+}
+
+u64 DeviceSession::GetPlayedSampleCount() const {
+    if (stream) {
+        return stream->GetPlayedSampleCount();
+    }
+    return 0;
+}
+
+} // namespace AudioCore

src/audio_core/device/device_session.h

@@ -0,0 +1,124 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <span>
+
+#include "audio_core/common/common.h"
+#include "audio_core/sink/sink.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace Core {
+class System;
+}
+
+namespace AudioCore {
+namespace Sink {
+class SinkStream;
+struct SinkBuffer;
+} // namespace Sink
+
+struct AudioBuffer;
+
+/**
+ * Represents an input or output device stream for audio in and audio out (not used for render).
+ **/
+class DeviceSession {
+public:
+    explicit DeviceSession(Core::System& system);
+    ~DeviceSession();
+
+    /**
+     * Initialize this device session.
+     *
+     * @param name                    - Name of this device.
+     * @param sample_format           - Sample format for this device's output.
+     * @param channel_count           - Number of channels for this device (2 or 6).
+     * @param session_id              - This session's id.
+     * @param handle                  - Handle for this device session (unused).
+     * @param applet_resource_user_id - Applet resource user id for this device session (unused).
+     * @param type                    - Type of this stream (Render, In, Out).
+     * @return Result code for this call.
+     */
+    Result Initialize(std::string_view name, SampleFormat sample_format, u16 channel_count,
+                      size_t session_id, u32 handle, u64 applet_resource_user_id,
+                      Sink::StreamType type);
+
+    /**
+     * Finalize this device session.
+     */
+    void Finalize();
+
+    /**
+     * Append audio buffers to this device session to be played back.
+     *
+     * @param buffers - The buffers to play.
+     */
+    void AppendBuffers(std::span<AudioBuffer> buffers) const;
+
+    /**
+     * (Audio In only) Pop samples from the backend, and write them back to this buffer's address.
+     *
+     * @param buffer - The buffer to write to.
+     */
+    void ReleaseBuffer(AudioBuffer& buffer) const;
+
+    /**
+     * Check if the buffer for the given tag has been consumed by the backend.
+     *
+     * @param tag - Unqiue tag of the buffer to check.
+     * @return true if the buffer has been consumed, otherwise false.
+     */
+    bool IsBufferConsumed(u64 tag) const;
+
+    /**
+     * Start this device session, starting the backend stream.
+     */
+    void Start();
+
+    /**
+     * Stop this device session, stopping the backend stream.
+     */
+    void Stop();
+
+    /**
+     * Set this device session's volume.
+     *
+     * @param volume - New volume for this session.
+     */
+    void SetVolume(f32 volume) const;
+
+    /**
+     * Get this device session's total played sample count.
+     *
+     * @return Samples played by this session.
+     */
+    u64 GetPlayedSampleCount() const;
+
+private:
+    /// System
+    Core::System& system;
+    /// Output sink this device will use
+    Sink::Sink* sink;
+    /// The backend stream for this device session to send samples to
+    Sink::SinkStream* stream{};
+    /// Name of this device session
+    std::string name{};
+    /// Type of this device session (render/in/out)
+    Sink::StreamType type{};
+    /// Sample format for this device.
+    SampleFormat sample_format{SampleFormat::PcmInt16};
+    /// Channel count for this device session
+    u16 channel_count{};
+    /// Session id of this device session
+    size_t session_id{};
+    /// Handle of this device session
+    u32 handle{};
+    /// Applet resource user id of this device session
+    u64 applet_resource_user_id{};
+    /// Total number of samples played by this device session
+    u64 played_sample_count{};
+};
+
+} // namespace AudioCore

src/audio_core/in/audio_in.cpp

@@ -0,0 +1,100 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_in_manager.h"
+#include "audio_core/in/audio_in.h"
+#include "core/hle/kernel/k_event.h"
+
+namespace AudioCore::AudioIn {
+
+In::In(Core::System& system_, Manager& manager_, Kernel::KEvent* event_, size_t session_id_)
+    : manager{manager_}, parent_mutex{manager.mutex}, event{event_}, system{system_, event,
+                                                                            session_id_} {}
+
+void In::Free() {
+    std::scoped_lock l{parent_mutex};
+    manager.ReleaseSessionId(system.GetSessionId());
+}
+
+System& In::GetSystem() {
+    return system;
+}
+
+AudioIn::State In::GetState() {
+    std::scoped_lock l{parent_mutex};
+    return system.GetState();
+}
+
+Result In::StartSystem() {
+    std::scoped_lock l{parent_mutex};
+    return system.Start();
+}
+
+void In::StartSession() {
+    std::scoped_lock l{parent_mutex};
+    system.StartSession();
+}
+
+Result In::StopSystem() {
+    std::scoped_lock l{parent_mutex};
+    return system.Stop();
+}
+
+Result In::AppendBuffer(const AudioInBuffer& buffer, u64 tag) {
+    std::scoped_lock l{parent_mutex};
+
+    if (system.AppendBuffer(buffer, tag)) {
+        return ResultSuccess;
+    }
+    return Service::Audio::ERR_BUFFER_COUNT_EXCEEDED;
+}
+
+void In::ReleaseAndRegisterBuffers() {
+    std::scoped_lock l{parent_mutex};
+    if (system.GetState() == State::Started) {
+        system.ReleaseBuffers();
+        system.RegisterBuffers();
+    }
+}
+
+bool In::FlushAudioInBuffers() {
+    std::scoped_lock l{parent_mutex};
+    return system.FlushAudioInBuffers();
+}
+
+u32 In::GetReleasedBuffers(std::span<u64> tags) {
+    std::scoped_lock l{parent_mutex};
+    return system.GetReleasedBuffers(tags);
+}
+
+Kernel::KReadableEvent& In::GetBufferEvent() {
+    std::scoped_lock l{parent_mutex};
+    return event->GetReadableEvent();
+}
+
+f32 In::GetVolume() {
+    std::scoped_lock l{parent_mutex};
+    return system.GetVolume();
+}
+
+void In::SetVolume(f32 volume) {
+    std::scoped_lock l{parent_mutex};
+    system.SetVolume(volume);
+}
+
+bool In::ContainsAudioBuffer(u64 tag) {
+    std::scoped_lock l{parent_mutex};
+    return system.ContainsAudioBuffer(tag);
+}
+
+u32 In::GetBufferCount() {
+    std::scoped_lock l{parent_mutex};
+    return system.GetBufferCount();
+}
+
+u64 In::GetPlayedSampleCount() {
+    std::scoped_lock l{parent_mutex};
+    return system.GetPlayedSampleCount();
+}
+
+} // namespace AudioCore::AudioIn

src/audio_core/in/audio_in.h

@@ -0,0 +1,147 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <mutex>
+
+#include "audio_core/in/audio_in_system.h"
+
+namespace Core {
+class System;
+}
+
+namespace Kernel {
+class KEvent;
+class KReadableEvent;
+} // namespace Kernel
+
+namespace AudioCore::AudioIn {
+class Manager;
+
+/**
+ * Interface between the service and audio in system. Mainly responsible for forwarding service
+ * calls to the system.
+ */
+class In {
+public:
+    explicit In(Core::System& system, Manager& manager, Kernel::KEvent* event, size_t session_id);
+
+    /**
+     * Free this audio in from the audio in manager.
+     */
+    void Free();
+
+    /**
+     * Get this audio in's system.
+     */
+    System& GetSystem();
+
+    /**
+     * Get the current state.
+     *
+     * @return Started or Stopped.
+     */
+    AudioIn::State GetState();
+
+    /**
+     * Start the system
+     *
+     * @return Result code
+     */
+    Result StartSystem();
+
+    /**
+     * Start the system's device session.
+     */
+    void StartSession();
+
+    /**
+     * Stop the system.
+     *
+     * @return Result code
+     */
+    Result StopSystem();
+
+    /**
+     * Append a new buffer to the system, the buffer event will be signalled when it is filled.
+     *
+     * @param buffer - The new buffer to append.
+     * @param tag    - Unique tag for this buffer.
+     * @return Result code.
+     */
+    Result AppendBuffer(const AudioInBuffer& buffer, u64 tag);
+
+    /**
+     * Release all completed buffers, and register any appended.
+     */
+    void ReleaseAndRegisterBuffers();
+
+    /**
+     * Flush all buffers.
+     */
+    bool FlushAudioInBuffers();
+
+    /**
+     * Get all of the currently released buffers.
+     *
+     * @param tags - Output container for the buffer tags which were released.
+     * @return The number of buffers released.
+     */
+    u32 GetReleasedBuffers(std::span<u64> tags);
+
+    /**
+     * Get the buffer event for this audio in, this event will be signalled when a buffer is filled.
+     *
+     * @return The buffer event.
+     */
+    Kernel::KReadableEvent& GetBufferEvent();
+
+    /**
+     * Get the current system volume.
+     *
+     * @return The current volume.
+     */
+    f32 GetVolume();
+
+    /**
+     * Set the system volume.
+     *
+     * @param volume - The volume to set.
+     */
+    void SetVolume(f32 volume);
+
+    /**
+     * Check if a buffer is in the system.
+     *
+     * @param tag - The tag to search for.
+     * @return True if the buffer is in the system, otherwise false.
+     */
+    bool ContainsAudioBuffer(u64 tag);
+
+    /**
+     * Get the maximum number of buffers.
+     *
+     * @return The maximum number of buffers.
+     */
+    u32 GetBufferCount();
+
+    /**
+     * Get the total played sample count for this audio in.
+     *
+     * @return The played sample count.
+     */
+    u64 GetPlayedSampleCount();
+
+private:
+    /// The AudioIn::Manager this audio in is registered with
+    Manager& manager;
+    /// Manager's mutex
+    std::recursive_mutex& parent_mutex;
+    /// Buffer event, signalled when buffers are ready to be released
+    Kernel::KEvent* event;
+    /// Main audio in system
+    System system;
+};
+
+} // namespace AudioCore::AudioIn

src/audio_core/in/audio_in_system.cpp

@@ -0,0 +1,213 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <mutex>
+#include "audio_core/audio_event.h"
+#include "audio_core/audio_manager.h"
+#include "audio_core/in/audio_in_system.h"
+#include "common/logging/log.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/hle/kernel/k_event.h"
+
+namespace AudioCore::AudioIn {
+
+System::System(Core::System& system_, Kernel::KEvent* event_, const size_t session_id_)
+    : system{system_}, buffer_event{event_},
+      session_id{session_id_}, session{std::make_unique<DeviceSession>(system_)} {}
+
+System::~System() {
+    Finalize();
+}
+
+void System::Finalize() {
+    Stop();
+    session->Finalize();
+    buffer_event->GetWritableEvent().Signal();
+}
+
+void System::StartSession() {
+    session->Start();
+}
+
+size_t System::GetSessionId() const {
+    return session_id;
+}
+
+std::string_view System::GetDefaultDeviceName() {
+    return "BuiltInHeadset";
+}
+
+std::string_view System::GetDefaultUacDeviceName() {
+    return "Uac";
+}
+
+Result System::IsConfigValid(const std::string_view device_name,
+                             const AudioInParameter& in_params) {
+    if ((device_name.size() > 0) &&
+        (device_name != GetDefaultDeviceName() && device_name != GetDefaultUacDeviceName())) {
+        return Service::Audio::ERR_INVALID_DEVICE_NAME;
+    }
+
+    if (in_params.sample_rate != TargetSampleRate && in_params.sample_rate > 0) {
+        return Service::Audio::ERR_INVALID_SAMPLE_RATE;
+    }
+
+    return ResultSuccess;
+}
+
+Result System::Initialize(std::string& device_name, const AudioInParameter& in_params,
+                          const u32 handle_, const u64 applet_resource_user_id_) {
+    auto result{IsConfigValid(device_name, in_params)};
+    if (result.IsError()) {
+        return result;
+    }
+
+    handle = handle_;
+    applet_resource_user_id = applet_resource_user_id_;
+    if (device_name.empty() || device_name[0] == '\0') {
+        name = std::string(GetDefaultDeviceName());
+    } else {
+        name = std::move(device_name);
+    }
+
+    sample_rate = TargetSampleRate;
+    sample_format = SampleFormat::PcmInt16;
+    channel_count = in_params.channel_count <= 2 ? 2 : 6;
+    volume = 1.0f;
+    is_uac = name == "Uac";
+    return ResultSuccess;
+}
+
+Result System::Start() {
+    if (state != State::Stopped) {
+        return Service::Audio::ERR_OPERATION_FAILED;
+    }
+
+    session->Initialize(name, sample_format, channel_count, session_id, handle,
+                        applet_resource_user_id, Sink::StreamType::In);
+    session->SetVolume(volume);
+    session->Start();
+    state = State::Started;
+
+    std::vector<AudioBuffer> buffers_to_flush{};
+    buffers.RegisterBuffers(buffers_to_flush);
+    session->AppendBuffers(buffers_to_flush);
+
+    return ResultSuccess;
+}
+
+Result System::Stop() {
+    if (state == State::Started) {
+        session->Stop();
+        session->SetVolume(0.0f);
+        state = State::Stopped;
+    }
+
+    return ResultSuccess;
+}
+
+bool System::AppendBuffer(const AudioInBuffer& buffer, const u64 tag) {
+    if (buffers.GetTotalBufferCount() == BufferCount) {
+        return false;
+    }
+
+    AudioBuffer new_buffer{
+        .played_timestamp = 0, .samples = buffer.samples, .tag = tag, .size = buffer.size};
+
+    buffers.AppendBuffer(new_buffer);
+    RegisterBuffers();
+
+    return true;
+}
+
+void System::RegisterBuffers() {
+    if (state == State::Started) {
+        std::vector<AudioBuffer> registered_buffers{};
+        buffers.RegisterBuffers(registered_buffers);
+        session->AppendBuffers(registered_buffers);
+    }
+}
+
+void System::ReleaseBuffers() {
+    bool signal{buffers.ReleaseBuffers(system.CoreTiming(), *session)};
+
+    if (signal) {
+        // Signal if any buffer was released, or if none are registered, we need more.
+        buffer_event->GetWritableEvent().Signal();
+    }
+}
+
+u32 System::GetReleasedBuffers(std::span<u64> tags) {
+    return buffers.GetReleasedBuffers(tags);
+}
+
+bool System::FlushAudioInBuffers() {
+    if (state != State::Started) {
+        return false;
+    }
+
+    u32 buffers_released{};
+    buffers.FlushBuffers(buffers_released);
+
+    if (buffers_released > 0) {
+        buffer_event->GetWritableEvent().Signal();
+    }
+    return true;
+}
+
+u16 System::GetChannelCount() const {
+    return channel_count;
+}
+
+u32 System::GetSampleRate() const {
+    return sample_rate;
+}
+
+SampleFormat System::GetSampleFormat() const {
+    return sample_format;
+}
+
+State System::GetState() {
+    switch (state) {
+    case State::Started:
+    case State::Stopped:
+        return state;
+    default:
+        LOG_ERROR(Service_Audio, "AudioIn invalid state!");
+        state = State::Stopped;
+        break;
+    }
+    return state;
+}
+
+std::string System::GetName() const {
+    return name;
+}
+
+f32 System::GetVolume() const {
+    return volume;
+}
+
+void System::SetVolume(const f32 volume_) {
+    volume = volume_;
+    session->SetVolume(volume_);
+}
+
+bool System::ContainsAudioBuffer(const u64 tag) {
+    return buffers.ContainsBuffer(tag);
+}
+
+u32 System::GetBufferCount() {
+    return buffers.GetAppendedRegisteredCount();
+}
+
+u64 System::GetPlayedSampleCount() const {
+    return session->GetPlayedSampleCount();
+}
+
+bool System::IsUac() const {
+    return is_uac;
+}
+
+} // namespace AudioCore::AudioIn

src/audio_core/in/audio_in_system.h

@@ -0,0 +1,275 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <atomic>
+#include <memory>
+#include <span>
+#include <string>
+
+#include "audio_core/common/common.h"
+#include "audio_core/device/audio_buffers.h"
+#include "audio_core/device/device_session.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace Core {
+class System;
+}
+
+namespace Kernel {
+class KEvent;
+}
+
+namespace AudioCore::AudioIn {
+
+constexpr SessionTypes SessionType = SessionTypes::AudioIn;
+
+struct AudioInParameter {
+    /* 0x0 */ s32_le sample_rate;
+    /* 0x4 */ u16_le channel_count;
+    /* 0x6 */ u16_le reserved;
+};
+static_assert(sizeof(AudioInParameter) == 0x8, "AudioInParameter is an invalid size");
+
+struct AudioInParameterInternal {
+    /* 0x0 */ u32_le sample_rate;
+    /* 0x4 */ u32_le channel_count;
+    /* 0x8 */ u32_le sample_format;
+    /* 0xC */ u32_le state;
+};
+static_assert(sizeof(AudioInParameterInternal) == 0x10,
+              "AudioInParameterInternal is an invalid size");
+
+struct AudioInBuffer {
+    /* 0x00 */ AudioInBuffer* next;
+    /* 0x08 */ VAddr samples;
+    /* 0x10 */ u64 capacity;
+    /* 0x18 */ u64 size;
+    /* 0x20 */ u64 offset;
+};
+static_assert(sizeof(AudioInBuffer) == 0x28, "AudioInBuffer is an invalid size");
+
+enum class State {
+    Started,
+    Stopped,
+};
+
+/**
+ * Controls and drives audio input.
+ */
+class System {
+public:
+    explicit System(Core::System& system, Kernel::KEvent* event, size_t session_id);
+    ~System();
+
+    /**
+     * Get the default audio input device name.
+     *
+     * @return The default audio input device name.
+     */
+    std::string_view GetDefaultDeviceName();
+
+    /**
+     * Get the default USB audio input device name.
+     * This is preferred over non-USB as some games refuse to work with the BuiltInHeadset
+     * (e.g Let's Sing).
+     *
+     * @return The default USB audio input device name.
+     */
+    std::string_view GetDefaultUacDeviceName();
+
+    /**
+     * Is the given initialize config valid?
+     *
+     * @param device_name - The name of the requested input device.
+     * @param in_params   - Input parameters, see AudioInParameter.
+     * @return Result code.
+     */
+    Result IsConfigValid(std::string_view device_name, const AudioInParameter& in_params);
+
+    /**
+     * Initialize this system.
+     *
+     * @param device_name             - The name of the requested input device.
+     * @param in_params               - Input parameters, see AudioInParameter.
+     * @param handle                  - Unused.
+     * @param applet_resource_user_id - Unused.
+     * @return Result code.
+     */
+    Result Initialize(std::string& device_name, const AudioInParameter& in_params, u32 handle,
+                      u64 applet_resource_user_id);
+
+    /**
+     * Start this system.
+     *
+     * @return Result code.
+     */
+    Result Start();
+
+    /**
+     * Stop this system.
+     *
+     * @return Result code.
+     */
+    Result Stop();
+
+    /**
+     * Finalize this system.
+     */
+    void Finalize();
+
+    /**
+     * Start this system's device session.
+     */
+    void StartSession();
+
+    /**
+     * Get this system's id.
+     */
+    size_t GetSessionId() const;
+
+    /**
+     * Append a new buffer to the device.
+     *
+     * @param buffer - New buffer to append.
+     * @param tag    - Unique tag of the buffer.
+     * @return True if the buffer was appended, otherwise false.
+     */
+    bool AppendBuffer(const AudioInBuffer& buffer, u64 tag);
+
+    /**
+     * Register all appended buffers.
+     */
+    void RegisterBuffers();
+
+    /**
+     * Release all registered buffers.
+     */
+    void ReleaseBuffers();
+
+    /**
+     * Get all released buffers.
+     *
+     * @param tags - Container to be filled with the released buffers' tags.
+     * @return The number of buffers released.
+     */
+    u32 GetReleasedBuffers(std::span<u64> tags);
+
+    /**
+     * Flush all appended and registered buffers.
+     *
+     * @return True if buffers were successfully flushed, otherwise false.
+     */
+    bool FlushAudioInBuffers();
+
+    /**
+     * Get this system's current channel count.
+     *
+     * @return The channel count.
+     */
+    u16 GetChannelCount() const;
+
+    /**
+     * Get this system's current sample rate.
+     *
+     * @return The sample rate.
+     */
+    u32 GetSampleRate() const;
+
+    /**
+     * Get this system's current sample format.
+     *
+     * @return The sample format.
+     */
+    SampleFormat GetSampleFormat() const;
+
+    /**
+     * Get this system's current state.
+     *
+     * @return The current state.
+     */
+    State GetState();
+
+    /**
+     * Get this system's name.
+     *
+     * @return The system's name.
+     */
+    std::string GetName() const;
+
+    /**
+     * Get this system's current volume.
+     *
+     * @return The system's current volume.
+     */
+    f32 GetVolume() const;
+
+    /**
+     * Set this system's current volume.
+     *
+     * @param The new volume.
+     */
+    void SetVolume(f32 volume);
+
+    /**
+     * Does the system contain this buffer?
+     *
+     * @param tag - Unique tag to search for.
+     * @return True if the buffer is in the system, otherwise false.
+     */
+    bool ContainsAudioBuffer(u64 tag);
+
+    /**
+     * Get the maximum number of usable buffers (default 32).
+     *
+     * @return The number of buffers.
+     */
+    u32 GetBufferCount();
+
+    /**
+     * Get the total number of samples played by this system.
+     *
+     * @return The number of samples.
+     */
+    u64 GetPlayedSampleCount() const;
+
+    /**
+     * Is this system using a USB device?
+     *
+     * @return True if using a USB device, otherwise false.
+     */
+    bool IsUac() const;
+
+private:
+    /// Core system
+    Core::System& system;
+    /// (Unused)
+    u32 handle{};
+    /// (Unused)
+    u64 applet_resource_user_id{};
+    /// Buffer event, signalled when a buffer is ready
+    Kernel::KEvent* buffer_event;
+    /// Session id of this system
+    size_t session_id{};
+    /// Device session for this system
+    std::unique_ptr<DeviceSession> session;
+    /// Audio buffers in use by this system
+    AudioBuffers<BufferCount> buffers{BufferCount};
+    /// Sample rate of this system
+    u32 sample_rate{};
+    /// Sample format of this system
+    SampleFormat sample_format{SampleFormat::PcmInt16};
+    /// Channel count of this system
+    u16 channel_count{};
+    /// State of this system
+    std::atomic<State> state{State::Stopped};
+    /// Name of this system
+    std::string name{};
+    /// Volume of this system
+    f32 volume{1.0f};
+    /// Is this system's device USB?
+    bool is_uac{false};
+};
+
+} // namespace AudioCore::AudioIn

src/audio_core/out/audio_out.cpp

@@ -0,0 +1,100 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_out_manager.h"
+#include "audio_core/out/audio_out.h"
+#include "core/hle/kernel/k_event.h"
+
+namespace AudioCore::AudioOut {
+
+Out::Out(Core::System& system_, Manager& manager_, Kernel::KEvent* event_, size_t session_id_)
+    : manager{manager_}, parent_mutex{manager.mutex}, event{event_}, system{system_, event,
+                                                                            session_id_} {}
+
+void Out::Free() {
+    std::scoped_lock l{parent_mutex};
+    manager.ReleaseSessionId(system.GetSessionId());
+}
+
+System& Out::GetSystem() {
+    return system;
+}
+
+AudioOut::State Out::GetState() {
+    std::scoped_lock l{parent_mutex};
+    return system.GetState();
+}
+
+Result Out::StartSystem() {
+    std::scoped_lock l{parent_mutex};
+    return system.Start();
+}
+
+void Out::StartSession() {
+    std::scoped_lock l{parent_mutex};
+    system.StartSession();
+}
+
+Result Out::StopSystem() {
+    std::scoped_lock l{parent_mutex};
+    return system.Stop();
+}
+
+Result Out::AppendBuffer(const AudioOutBuffer& buffer, const u64 tag) {
+    std::scoped_lock l{parent_mutex};
+
+    if (system.AppendBuffer(buffer, tag)) {
+        return ResultSuccess;
+    }
+    return Service::Audio::ERR_BUFFER_COUNT_EXCEEDED;
+}
+
+void Out::ReleaseAndRegisterBuffers() {
+    std::scoped_lock l{parent_mutex};
+    if (system.GetState() == State::Started) {
+        system.ReleaseBuffers();
+        system.RegisterBuffers();
+    }
+}
+
+bool Out::FlushAudioOutBuffers() {
+    std::scoped_lock l{parent_mutex};
+    return system.FlushAudioOutBuffers();
+}
+
+u32 Out::GetReleasedBuffers(std::span<u64> tags) {
+    std::scoped_lock l{parent_mutex};
+    return system.GetReleasedBuffers(tags);
+}
+
+Kernel::KReadableEvent& Out::GetBufferEvent() {
+    std::scoped_lock l{parent_mutex};
+    return event->GetReadableEvent();
+}
+
+f32 Out::GetVolume() {
+    std::scoped_lock l{parent_mutex};
+    return system.GetVolume();
+}
+
+void Out::SetVolume(const f32 volume) {
+    std::scoped_lock l{parent_mutex};
+    system.SetVolume(volume);
+}
+
+bool Out::ContainsAudioBuffer(const u64 tag) {
+    std::scoped_lock l{parent_mutex};
+    return system.ContainsAudioBuffer(tag);
+}
+
+u32 Out::GetBufferCount() {
+    std::scoped_lock l{parent_mutex};
+    return system.GetBufferCount();
+}
+
+u64 Out::GetPlayedSampleCount() {
+    std::scoped_lock l{parent_mutex};
+    return system.GetPlayedSampleCount();
+}
+
+} // namespace AudioCore::AudioOut

src/audio_core/out/audio_out.h

@@ -0,0 +1,147 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <mutex>
+
+#include "audio_core/out/audio_out_system.h"
+
+namespace Core {
+class System;
+}
+
+namespace Kernel {
+class KEvent;
+class KReadableEvent;
+} // namespace Kernel
+
+namespace AudioCore::AudioOut {
+class Manager;
+
+/**
+ * Interface between the service and audio out system. Mainly responsible for forwarding service
+ * calls to the system.
+ */
+class Out {
+public:
+    explicit Out(Core::System& system, Manager& manager, Kernel::KEvent* event, size_t session_id);
+
+    /**
+     * Free this audio out from the audio out manager.
+     */
+    void Free();
+
+    /**
+     * Get this audio out's system.
+     */
+    System& GetSystem();
+
+    /**
+     * Get the current state.
+     *
+     * @return Started or Stopped.
+     */
+    AudioOut::State GetState();
+
+    /**
+     * Start the system
+     *
+     * @return Result code
+     */
+    Result StartSystem();
+
+    /**
+     * Start the system's device session.
+     */
+    void StartSession();
+
+    /**
+     * Stop the system.
+     *
+     * @return Result code
+     */
+    Result StopSystem();
+
+    /**
+     * Append a new buffer to the system, the buffer event will be signalled when it is filled.
+     *
+     * @param buffer - The new buffer to append.
+     * @param tag    - Unique tag for this buffer.
+     * @return Result code.
+     */
+    Result AppendBuffer(const AudioOutBuffer& buffer, u64 tag);
+
+    /**
+     * Release all completed buffers, and register any appended.
+     */
+    void ReleaseAndRegisterBuffers();
+
+    /**
+     * Flush all buffers.
+     */
+    bool FlushAudioOutBuffers();
+
+    /**
+     * Get all of the currently released buffers.
+     *
+     * @param tags - Output container for the buffer tags which were released.
+     * @return The number of buffers released.
+     */
+    u32 GetReleasedBuffers(std::span<u64> tags);
+
+    /**
+     * Get the buffer event for this audio out, this event will be signalled when a buffer is
+     * filled.
+     * @return The buffer event.
+     */
+    Kernel::KReadableEvent& GetBufferEvent();
+
+    /**
+     * Get the current system volume.
+     *
+     * @return The current volume.
+     */
+    f32 GetVolume();
+
+    /**
+     * Set the system volume.
+     *
+     * @param volume - The volume to set.
+     */
+    void SetVolume(f32 volume);
+
+    /**
+     * Check if a buffer is in the system.
+     *
+     * @param tag - The tag to search for.
+     * @return True if the buffer is in the system, otherwise false.
+     */
+    bool ContainsAudioBuffer(u64 tag);
+
+    /**
+     * Get the maximum number of buffers.
+     *
+     * @return The maximum number of buffers.
+     */
+    u32 GetBufferCount();
+
+    /**
+     * Get the total played sample count for this audio out.
+     *
+     * @return The played sample count.
+     */
+    u64 GetPlayedSampleCount();
+
+private:
+    /// The AudioOut::Manager this audio out is registered with
+    Manager& manager;
+    /// Manager's mutex
+    std::recursive_mutex& parent_mutex;
+    /// Buffer event, signalled when buffers are ready to be released
+    Kernel::KEvent* event;
+    /// Main audio out system
+    System system;
+};
+
+} // namespace AudioCore::AudioOut

src/audio_core/out/audio_out_system.cpp

@@ -0,0 +1,207 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <mutex>
+
+#include "audio_core/audio_event.h"
+#include "audio_core/audio_manager.h"
+#include "audio_core/out/audio_out_system.h"
+#include "common/logging/log.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/hle/kernel/k_event.h"
+
+namespace AudioCore::AudioOut {
+
+System::System(Core::System& system_, Kernel::KEvent* event_, size_t session_id_)
+    : system{system_}, buffer_event{event_},
+      session_id{session_id_}, session{std::make_unique<DeviceSession>(system_)} {}
+
+System::~System() {
+    Finalize();
+}
+
+void System::Finalize() {
+    Stop();
+    session->Finalize();
+    buffer_event->GetWritableEvent().Signal();
+}
+
+std::string_view System::GetDefaultOutputDeviceName() {
+    return "DeviceOut";
+}
+
+Result System::IsConfigValid(std::string_view device_name, const AudioOutParameter& in_params) {
+    if ((device_name.size() > 0) && (device_name != GetDefaultOutputDeviceName())) {
+        return Service::Audio::ERR_INVALID_DEVICE_NAME;
+    }
+
+    if (in_params.sample_rate != TargetSampleRate && in_params.sample_rate > 0) {
+        return Service::Audio::ERR_INVALID_SAMPLE_RATE;
+    }
+
+    if (in_params.channel_count == 0 || in_params.channel_count == 2 ||
+        in_params.channel_count == 6) {
+        return ResultSuccess;
+    }
+
+    return Service::Audio::ERR_INVALID_CHANNEL_COUNT;
+}
+
+Result System::Initialize(std::string& device_name, const AudioOutParameter& in_params, u32 handle_,
+                          u64& applet_resource_user_id_) {
+    auto result = IsConfigValid(device_name, in_params);
+    if (result.IsError()) {
+        return result;
+    }
+
+    handle = handle_;
+    applet_resource_user_id = applet_resource_user_id_;
+    if (device_name.empty() || device_name[0] == '\0') {
+        name = std::string(GetDefaultOutputDeviceName());
+    } else {
+        name = std::move(device_name);
+    }
+
+    sample_rate = TargetSampleRate;
+    sample_format = SampleFormat::PcmInt16;
+    channel_count = in_params.channel_count <= 2 ? 2 : 6;
+    volume = 1.0f;
+    return ResultSuccess;
+}
+
+void System::StartSession() {
+    session->Start();
+}
+
+size_t System::GetSessionId() const {
+    return session_id;
+}
+
+Result System::Start() {
+    if (state != State::Stopped) {
+        return Service::Audio::ERR_OPERATION_FAILED;
+    }
+
+    session->Initialize(name, sample_format, channel_count, session_id, handle,
+                        applet_resource_user_id, Sink::StreamType::Out);
+    session->SetVolume(volume);
+    session->Start();
+    state = State::Started;
+
+    std::vector<AudioBuffer> buffers_to_flush{};
+    buffers.RegisterBuffers(buffers_to_flush);
+    session->AppendBuffers(buffers_to_flush);
+
+    return ResultSuccess;
+}
+
+Result System::Stop() {
+    if (state == State::Started) {
+        session->Stop();
+        session->SetVolume(0.0f);
+        state = State::Stopped;
+    }
+
+    return ResultSuccess;
+}
+
+bool System::AppendBuffer(const AudioOutBuffer& buffer, u64 tag) {
+    if (buffers.GetTotalBufferCount() == BufferCount) {
+        return false;
+    }
+
+    AudioBuffer new_buffer{
+        .played_timestamp = 0, .samples = buffer.samples, .tag = tag, .size = buffer.size};
+
+    buffers.AppendBuffer(new_buffer);
+    RegisterBuffers();
+
+    return true;
+}
+
+void System::RegisterBuffers() {
+    if (state == State::Started) {
+        std::vector<AudioBuffer> registered_buffers{};
+        buffers.RegisterBuffers(registered_buffers);
+        session->AppendBuffers(registered_buffers);
+    }
+}
+
+void System::ReleaseBuffers() {
+    bool signal{buffers.ReleaseBuffers(system.CoreTiming(), *session)};
+    if (signal) {
+        // Signal if any buffer was released, or if none are registered, we need more.
+        buffer_event->GetWritableEvent().Signal();
+    }
+}
+
+u32 System::GetReleasedBuffers(std::span<u64> tags) {
+    return buffers.GetReleasedBuffers(tags);
+}
+
+bool System::FlushAudioOutBuffers() {
+    if (state != State::Started) {
+        return false;
+    }
+
+    u32 buffers_released{};
+    buffers.FlushBuffers(buffers_released);
+
+    if (buffers_released > 0) {
+        buffer_event->GetWritableEvent().Signal();
+    }
+    return true;
+}
+
+u16 System::GetChannelCount() const {
+    return channel_count;
+}
+
+u32 System::GetSampleRate() const {
+    return sample_rate;
+}
+
+SampleFormat System::GetSampleFormat() const {
+    return sample_format;
+}
+
+State System::GetState() {
+    switch (state) {
+    case State::Started:
+    case State::Stopped:
+        return state;
+    default:
+        LOG_ERROR(Service_Audio, "AudioOut invalid state!");
+        state = State::Stopped;
+        break;
+    }
+    return state;
+}
+
+std::string System::GetName() const {
+    return name;
+}
+
+f32 System::GetVolume() const {
+    return volume;
+}
+
+void System::SetVolume(const f32 volume_) {
+    volume = volume_;
+    session->SetVolume(volume_);
+}
+
+bool System::ContainsAudioBuffer(const u64 tag) {
+    return buffers.ContainsBuffer(tag);
+}
+
+u32 System::GetBufferCount() {
+    return buffers.GetAppendedRegisteredCount();
+}
+
+u64 System::GetPlayedSampleCount() const {
+    return session->GetPlayedSampleCount();
+}
+
+} // namespace AudioCore::AudioOut

src/audio_core/out/audio_out_system.h

@@ -0,0 +1,257 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <atomic>
+#include <memory>
+#include <span>
+#include <string>
+
+#include "audio_core/common/common.h"
+#include "audio_core/device/audio_buffers.h"
+#include "audio_core/device/device_session.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace Core {
+class System;
+}
+
+namespace Kernel {
+class KEvent;
+}
+
+namespace AudioCore::AudioOut {
+
+constexpr SessionTypes SessionType = SessionTypes::AudioOut;
+
+struct AudioOutParameter {
+    /* 0x0 */ s32_le sample_rate;
+    /* 0x4 */ u16_le channel_count;
+    /* 0x6 */ u16_le reserved;
+};
+static_assert(sizeof(AudioOutParameter) == 0x8, "AudioOutParameter is an invalid size");
+
+struct AudioOutParameterInternal {
+    /* 0x0 */ u32_le sample_rate;
+    /* 0x4 */ u32_le channel_count;
+    /* 0x8 */ u32_le sample_format;
+    /* 0xC */ u32_le state;
+};
+static_assert(sizeof(AudioOutParameterInternal) == 0x10,
+              "AudioOutParameterInternal is an invalid size");
+
+struct AudioOutBuffer {
+    /* 0x00 */ AudioOutBuffer* next;
+    /* 0x08 */ VAddr samples;
+    /* 0x10 */ u64 capacity;
+    /* 0x18 */ u64 size;
+    /* 0x20 */ u64 offset;
+};
+static_assert(sizeof(AudioOutBuffer) == 0x28, "AudioOutBuffer is an invalid size");
+
+enum class State {
+    Started,
+    Stopped,
+};
+
+/**
+ * Controls and drives audio output.
+ */
+class System {
+public:
+    explicit System(Core::System& system, Kernel::KEvent* event, size_t session_id);
+    ~System();
+
+    /**
+     * Get the default audio output device name.
+     *
+     * @return The default audio output device name.
+     */
+    std::string_view GetDefaultOutputDeviceName();
+
+    /**
+     * Is the given initialize config valid?
+     *
+     * @param device_name - The name of the requested output device.
+     * @param in_params   - Input parameters, see AudioOutParameter.
+     * @return Result code.
+     */
+    Result IsConfigValid(std::string_view device_name, const AudioOutParameter& in_params);
+
+    /**
+     * Initialize this system.
+     *
+     * @param device_name             - The name of the requested output device.
+     * @param in_params               - Input parameters, see AudioOutParameter.
+     * @param handle                  - Unused.
+     * @param applet_resource_user_id - Unused.
+     * @return Result code.
+     */
+    Result Initialize(std::string& device_name, const AudioOutParameter& in_params, u32 handle,
+                      u64& applet_resource_user_id);
+
+    /**
+     * Start this system.
+     *
+     * @return Result code.
+     */
+    Result Start();
+
+    /**
+     * Stop this system.
+     *
+     * @return Result code.
+     */
+    Result Stop();
+
+    /**
+     * Finalize this system.
+     */
+    void Finalize();
+
+    /**
+     * Start this system's device session.
+     */
+    void StartSession();
+
+    /**
+     * Get this system's id.
+     */
+    size_t GetSessionId() const;
+
+    /**
+     * Append a new buffer to the device.
+     *
+     * @param buffer - New buffer to append.
+     * @param tag    - Unique tag of the buffer.
+     * @return True if the buffer was appended, otherwise false.
+     */
+    bool AppendBuffer(const AudioOutBuffer& buffer, u64 tag);
+
+    /**
+     * Register all appended buffers.
+     */
+    void RegisterBuffers();
+
+    /**
+     * Release all registered buffers.
+     */
+    void ReleaseBuffers();
+
+    /**
+     * Get all released buffers.
+     *
+     * @param tags - Container to be filled with the released buffers' tags.
+     * @return The number of buffers released.
+     */
+    u32 GetReleasedBuffers(std::span<u64> tags);
+
+    /**
+     * Flush all appended and registered buffers.
+     *
+     * @return True if buffers were successfully flushed, otherwise false.
+     */
+    bool FlushAudioOutBuffers();
+
+    /**
+     * Get this system's current channel count.
+     *
+     * @return The channel count.
+     */
+    u16 GetChannelCount() const;
+
+    /**
+     * Get this system's current sample rate.
+     *
+     * @return The sample rate.
+     */
+    u32 GetSampleRate() const;
+
+    /**
+     * Get this system's current sample format.
+     *
+     * @return The sample format.
+     */
+    SampleFormat GetSampleFormat() const;
+
+    /**
+     * Get this system's current state.
+     *
+     * @return The current state.
+     */
+    State GetState();
+
+    /**
+     * Get this system's name.
+     *
+     * @return The system's name.
+     */
+    std::string GetName() const;
+
+    /**
+     * Get this system's current volume.
+     *
+     * @return The system's current volume.
+     */
+    f32 GetVolume() const;
+
+    /**
+     * Set this system's current volume.
+     *
+     * @param The new volume.
+     */
+    void SetVolume(f32 volume);
+
+    /**
+     * Does the system contain this buffer?
+     *
+     * @param tag - Unique tag to search for.
+     * @return True if the buffer is in the system, otherwise false.
+     */
+    bool ContainsAudioBuffer(u64 tag);
+
+    /**
+     * Get the maximum number of usable buffers (default 32).
+     *
+     * @return The number of buffers.
+     */
+    u32 GetBufferCount();
+
+    /**
+     * Get the total number of samples played by this system.
+     *
+     * @return The number of samples.
+     */
+    u64 GetPlayedSampleCount() const;
+
+private:
+    /// Core system
+    Core::System& system;
+    /// (Unused)
+    u32 handle{};
+    /// (Unused)
+    u64 applet_resource_user_id{};
+    /// Buffer event, signalled when a buffer is ready
+    Kernel::KEvent* buffer_event;
+    /// Session id of this system
+    size_t session_id{};
+    /// Device session for this system
+    std::unique_ptr<DeviceSession> session;
+    /// Audio buffers in use by this system
+    AudioBuffers<BufferCount> buffers{BufferCount};
+    /// Sample rate of this system
+    u32 sample_rate{};
+    /// Sample format of this system
+    SampleFormat sample_format{SampleFormat::PcmInt16};
+    /// Channel count of this system
+    u16 channel_count{};
+    /// State of this system
+    std::atomic<State> state{State::Stopped};
+    /// Name of this system
+    std::string name{};
+    /// Volume of this system
+    f32 volume{1.0f};
+};
+
+} // namespace AudioCore::AudioOut

src/audio_core/renderer/adsp/adsp.cpp

@@ -0,0 +1,118 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/adsp/adsp.h"
+#include "audio_core/renderer/adsp/command_buffer.h"
+#include "audio_core/sink/sink.h"
+#include "common/logging/log.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/core_timing_util.h"
+#include "core/memory.h"
+
+namespace AudioCore::AudioRenderer::ADSP {
+
+ADSP::ADSP(Core::System& system_, Sink::Sink& sink_)
+    : system{system_}, memory{system.Memory()}, sink{sink_} {}
+
+ADSP::~ADSP() {
+    ClearCommandBuffers();
+}
+
+State ADSP::GetState() const {
+    if (running) {
+        return State::Started;
+    }
+    return State::Stopped;
+}
+
+AudioRenderer_Mailbox* ADSP::GetRenderMailbox() {
+    return &render_mailbox;
+}
+
+void ADSP::ClearRemainCount(const u32 session_id) {
+    render_mailbox.ClearRemainCount(session_id);
+}
+
+u64 ADSP::GetSignalledTick() const {
+    return render_mailbox.GetSignalledTick();
+}
+
+u64 ADSP::GetTimeTaken() const {
+    return render_mailbox.GetRenderTimeTaken();
+}
+
+u64 ADSP::GetRenderTimeTaken(const u32 session_id) {
+    return render_mailbox.GetCommandBuffer(session_id).render_time_taken;
+}
+
+u32 ADSP::GetRemainCommandCount(const u32 session_id) const {
+    return render_mailbox.GetRemainCommandCount(session_id);
+}
+
+void ADSP::SendCommandBuffer(const u32 session_id, CommandBuffer& command_buffer) {
+    render_mailbox.SetCommandBuffer(session_id, command_buffer);
+}
+
+u64 ADSP::GetRenderingStartTick(const u32 session_id) {
+    return render_mailbox.GetSignalledTick() +
+           render_mailbox.GetCommandBuffer(session_id).render_time_taken;
+}
+
+bool ADSP::Start() {
+    if (running) {
+        return running;
+    }
+
+    running = true;
+    systems_active++;
+    audio_renderer = std::make_unique<AudioRenderer>(system);
+    audio_renderer->Start(&render_mailbox);
+    render_mailbox.HostSendMessage(RenderMessage::AudioRenderer_InitializeOK);
+    if (render_mailbox.HostWaitMessage() != RenderMessage::AudioRenderer_InitializeOK) {
+        LOG_ERROR(
+            Service_Audio,
+            "Host Audio Renderer -- Failed to receive initialize message response from ADSP!");
+    }
+    return running;
+}
+
+void ADSP::Stop() {
+    systems_active--;
+    if (running && systems_active == 0) {
+        {
+            std::scoped_lock l{mailbox_lock};
+            render_mailbox.HostSendMessage(RenderMessage::AudioRenderer_Shutdown);
+            if (render_mailbox.HostWaitMessage() != RenderMessage::AudioRenderer_Shutdown) {
+                LOG_ERROR(Service_Audio, "Host Audio Renderer -- Failed to receive shutdown "
+                                         "message response from ADSP!");
+            }
+        }
+        audio_renderer->Stop();
+        running = false;
+    }
+}
+
+void ADSP::Signal() {
+    const auto signalled_tick{system.CoreTiming().GetClockTicks()};
+    render_mailbox.SetSignalledTick(signalled_tick);
+    render_mailbox.HostSendMessage(RenderMessage::AudioRenderer_Render);
+}
+
+void ADSP::Wait() {
+    std::scoped_lock l{mailbox_lock};
+    auto response{render_mailbox.HostWaitMessage()};
+    if (response != RenderMessage::AudioRenderer_RenderResponse) {
+        LOG_ERROR(Service_Audio, "Invalid ADSP response message, expected 0x{:02X}, got 0x{:02X}",
+                  static_cast<u32>(RenderMessage::AudioRenderer_RenderResponse),
+                  static_cast<u32>(response));
+    }
+
+    ClearCommandBuffers();
+}
+
+void ADSP::ClearCommandBuffers() {
+    render_mailbox.ClearCommandBuffers();
+}
+
+} // namespace AudioCore::AudioRenderer::ADSP

src/audio_core/renderer/adsp/adsp.h

@@ -0,0 +1,173 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <memory>
+#include <mutex>
+
+#include "audio_core/renderer/adsp/audio_renderer.h"
+#include "common/common_types.h"
+
+namespace Core {
+namespace Memory {
+class Memory;
+}
+class System;
+} // namespace Core
+
+namespace AudioCore {
+namespace Sink {
+class Sink;
+}
+
+namespace AudioRenderer::ADSP {
+struct CommandBuffer;
+
+enum class State {
+    Started,
+    Stopped,
+};
+
+/**
+ * Represents the ADSP embedded within the audio sysmodule.
+ * This is a 32-bit Linux4Tegra kernel from nVidia, which is launched with the sysmodule on boot.
+ *
+ * The kernel will run apps you program for it, Nintendo have the following:
+ *
+ * Gmix - Responsible for mixing final audio and sending it out to hardware. This is last place all
+ *        audio samples end up, and we skip it entirely, since we have very different backends and
+ *        mixing is implicitly handled by the OS (but also due to lack of research/simplicity).
+ *
+ * AudioRenderer - Receives command lists generated by the audio render
+ *                 system, processes them, and sends the samples to Gmix.
+ *
+ * OpusDecoder - Contains libopus, and controls processing Opus audio and sends it to Gmix.
+ *               Not much research done here, TODO if needed.
+ *
+ * We only implement the AudioRenderer for now.
+ *
+ * Communication for the apps is done through mailboxes, and some shared memory.
+ */
+class ADSP {
+public:
+    explicit ADSP(Core::System& system, Sink::Sink& sink);
+    ~ADSP();
+
+    /**
+     * Start the ADSP.
+     *
+     * @return True if started or already running, otherwise false.
+     */
+    bool Start();
+
+    /**
+     * Stop the ADSP.
+     *
+     * @return True if started or already running, otherwise false.
+     */
+    void Stop();
+
+    /**
+     * Get the ADSP's state.
+     *
+     * @return Started or Stopped.
+     */
+    State GetState() const;
+
+    /**
+     * Get the AudioRenderer mailbox to communicate with it.
+     *
+     * @return The AudioRenderer mailbox.
+     */
+    AudioRenderer_Mailbox* GetRenderMailbox();
+
+    /**
+     * Get the tick the ADSP was signalled.
+     *
+     * @return The tick the ADSP was signalled.
+     */
+    u64 GetSignalledTick() const;
+
+    /**
+     * Get the total time it took for the ADSP to run the last command lists (both command lists).
+     *
+     * @return The tick the ADSP was signalled.
+     */
+    u64 GetTimeTaken() const;
+
+    /**
+     * Get the last time a given command list took to run.
+     *
+     * @param session_id - The session id to check (0 or 1).
+     * @return The time it took.
+     */
+    u64 GetRenderTimeTaken(u32 session_id);
+
+    /**
+     * Clear the remaining command count for a given session.
+     *
+     * @param session_id - The session id to check (0 or 1).
+     */
+    void ClearRemainCount(u32 session_id);
+
+    /**
+     * Get the remaining number of commands left to process for a command list.
+     *
+     * @param session_id - The session id to check (0 or 1).
+     * @return The number of commands remaining.
+     */
+    u32 GetRemainCommandCount(u32 session_id) const;
+
+    /**
+     * Get the last tick a command list started processing.
+     *
+     * @param session_id - The session id to check (0 or 1).
+     * @return The last tick the given command list started.
+     */
+    u64 GetRenderingStartTick(u32 session_id);
+
+    /**
+     * Set a command buffer to be processed.
+     *
+     * @param session_id     - The session id to check (0 or 1).
+     * @param command_buffer - The command buffer to process.
+     */
+    void SendCommandBuffer(u32 session_id, CommandBuffer& command_buffer);
+
+    /**
+     * Clear the command buffers (does not clear the time taken or the remaining command count)
+     */
+    void ClearCommandBuffers();
+
+    /**
+     * Signal the AudioRenderer to begin processing.
+     */
+    void Signal();
+
+    /**
+     * Wait for the AudioRenderer to finish processing.
+     */
+    void Wait();
+
+private:
+    /// Core system
+    Core::System& system;
+    /// Core memory
+    Core::Memory::Memory& memory;
+    /// Number of systems active, used to prevent accidental shutdowns
+    u8 systems_active{0};
+    /// ADSP running state
+    std::atomic<bool> running{false};
+    /// Output sink used by the ADSP
+    Sink::Sink& sink;
+    /// AudioRenderer app
+    std::unique_ptr<AudioRenderer> audio_renderer{};
+    /// Communication for the AudioRenderer
+    AudioRenderer_Mailbox render_mailbox{};
+    /// Mailbox lock ffor the render mailbox
+    std::mutex mailbox_lock;
+};
+
+} // namespace AudioRenderer::ADSP
+} // namespace AudioCore

src/audio_core/renderer/adsp/audio_renderer.cpp

@@ -0,0 +1,230 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <array>
+#include <chrono>
+
+#include "audio_core/audio_core.h"
+#include "audio_core/common/common.h"
+#include "audio_core/renderer/adsp/audio_renderer.h"
+#include "audio_core/sink/sink.h"
+#include "common/logging/log.h"
+#include "common/microprofile.h"
+#include "common/thread.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/core_timing_util.h"
+
+MICROPROFILE_DEFINE(Audio_Renderer, "Audio", "DSP", MP_RGB(60, 19, 97));
+
+namespace AudioCore::AudioRenderer::ADSP {
+
+void AudioRenderer_Mailbox::HostSendMessage(RenderMessage message_) {
+    adsp_messages.enqueue(message_);
+    adsp_event.Set();
+}
+
+RenderMessage AudioRenderer_Mailbox::HostWaitMessage() {
+    host_event.Wait();
+    RenderMessage msg{RenderMessage::Invalid};
+    if (!host_messages.try_dequeue(msg)) {
+        LOG_ERROR(Service_Audio, "Failed to dequeue host message!");
+    }
+    return msg;
+}
+
+void AudioRenderer_Mailbox::ADSPSendMessage(const RenderMessage message_) {
+    host_messages.enqueue(message_);
+    host_event.Set();
+}
+
+RenderMessage AudioRenderer_Mailbox::ADSPWaitMessage() {
+    adsp_event.Wait();
+    RenderMessage msg{RenderMessage::Invalid};
+    if (!adsp_messages.try_dequeue(msg)) {
+        LOG_ERROR(Service_Audio, "Failed to dequeue ADSP message!");
+    }
+    return msg;
+}
+
+CommandBuffer& AudioRenderer_Mailbox::GetCommandBuffer(const s32 session_id) {
+    return command_buffers[session_id];
+}
+
+void AudioRenderer_Mailbox::SetCommandBuffer(const u32 session_id, CommandBuffer& buffer) {
+    command_buffers[session_id] = buffer;
+}
+
+u64 AudioRenderer_Mailbox::GetRenderTimeTaken() const {
+    return command_buffers[0].render_time_taken + command_buffers[1].render_time_taken;
+}
+
+u64 AudioRenderer_Mailbox::GetSignalledTick() const {
+    return signalled_tick;
+}
+
+void AudioRenderer_Mailbox::SetSignalledTick(const u64 tick) {
+    signalled_tick = tick;
+}
+
+void AudioRenderer_Mailbox::ClearRemainCount(const u32 session_id) {
+    command_buffers[session_id].remaining_command_count = 0;
+}
+
+u32 AudioRenderer_Mailbox::GetRemainCommandCount(const u32 session_id) const {
+    return command_buffers[session_id].remaining_command_count;
+}
+
+void AudioRenderer_Mailbox::ClearCommandBuffers() {
+    command_buffers[0].buffer = 0;
+    command_buffers[0].size = 0;
+    command_buffers[0].reset_buffers = false;
+    command_buffers[1].buffer = 0;
+    command_buffers[1].size = 0;
+    command_buffers[1].reset_buffers = false;
+}
+
+AudioRenderer::AudioRenderer(Core::System& system_)
+    : system{system_}, sink{system.AudioCore().GetOutputSink()} {
+    CreateSinkStreams();
+}
+
+AudioRenderer::~AudioRenderer() {
+    Stop();
+    for (auto& stream : streams) {
+        if (stream) {
+            sink.CloseStream(stream);
+        }
+        stream = nullptr;
+    }
+}
+
+void AudioRenderer::Start(AudioRenderer_Mailbox* mailbox_) {
+    if (running) {
+        return;
+    }
+
+    mailbox = mailbox_;
+    thread = std::thread(&AudioRenderer::ThreadFunc, this);
+    for (auto& stream : streams) {
+        stream->Start();
+    }
+    running = true;
+}
+
+void AudioRenderer::Stop() {
+    if (!running) {
+        return;
+    }
+
+    for (auto& stream : streams) {
+        stream->Stop();
+    }
+    thread.join();
+    running = false;
+}
+
+void AudioRenderer::CreateSinkStreams() {
+    u32 channels{sink.GetDeviceChannels()};
+    for (u32 i = 0; i < MaxRendererSessions; i++) {
+        std::string name{fmt::format("ADSP_RenderStream-{}", i)};
+        streams[i] = sink.AcquireSinkStream(system, channels, name,
+                                            ::AudioCore::Sink::StreamType::Render, &render_event);
+        streams[i]->SetSystemChannels(streams[i]->GetDeviceChannels());
+    }
+}
+
+void AudioRenderer::ThreadFunc() {
+    constexpr char name[]{"yuzu:AudioRenderer"};
+    MicroProfileOnThreadCreate(name);
+    Common::SetCurrentThreadName(name);
+    Common::SetCurrentThreadPriority(Common::ThreadPriority::Critical);
+    if (mailbox->ADSPWaitMessage() != RenderMessage::AudioRenderer_InitializeOK) {
+        LOG_ERROR(Service_Audio,
+                  "ADSP Audio Renderer -- Failed to receive initialize message from host!");
+        return;
+    }
+
+    mailbox->ADSPSendMessage(RenderMessage::AudioRenderer_InitializeOK);
+
+    constexpr u64 max_process_time{2'304'000ULL};
+
+    while (true) {
+        auto message{mailbox->ADSPWaitMessage()};
+        switch (message) {
+        case RenderMessage::AudioRenderer_Shutdown:
+            mailbox->ADSPSendMessage(RenderMessage::AudioRenderer_Shutdown);
+            return;
+
+        case RenderMessage::AudioRenderer_Render: {
+            std::array<bool, MaxRendererSessions> buffers_reset{};
+            std::array<u64, MaxRendererSessions> render_times_taken{};
+            const auto start_time{system.CoreTiming().GetClockTicks()};
+
+            for (u32 index = 0; index < 2; index++) {
+                auto& command_buffer{mailbox->GetCommandBuffer(index)};
+                auto& command_list_processor{command_list_processors[index]};
+
+                // Check this buffer is valid, as it may not be used.
+                if (command_buffer.buffer != 0) {
+                    // If there are no remaining commands (from the previous list),
+                    // this is a new command list, initalize it.
+                    if (command_buffer.remaining_command_count == 0) {
+                        command_list_processor.Initialize(system, command_buffer.buffer,
+                                                          command_buffer.size, streams[index]);
+                    }
+
+                    if (command_buffer.reset_buffers && !buffers_reset[index]) {
+                        streams[index]->ClearQueue();
+                        buffers_reset[index] = true;
+                    }
+
+                    u64 max_time{max_process_time};
+                    if (index == 1 && command_buffer.applet_resource_user_id ==
+                                          mailbox->GetCommandBuffer(0).applet_resource_user_id) {
+                        max_time = max_process_time -
+                                   Core::Timing::CyclesToNs(render_times_taken[0]).count();
+                        if (render_times_taken[0] > max_process_time) {
+                            max_time = 0;
+                        }
+                    }
+
+                    max_time = std::min(command_buffer.time_limit, max_time);
+                    command_list_processor.SetProcessTimeMax(max_time);
+
+                    // Process the command list
+                    {
+                        MICROPROFILE_SCOPE(Audio_Renderer);
+                        render_times_taken[index] =
+                            command_list_processor.Process(index) - start_time;
+                    }
+
+                    // If the stream queue is building up too much, wait for a signal
+                    // from the backend that a buffer was consumed.
+                    // In practice this will wait longer than 1 buffer due to timing.
+                    auto stream{command_list_processor.GetOutputSinkStream()};
+                    if (stream->GetQueueSize() >= 4) {
+                        render_event.WaitFor(std::chrono::milliseconds(5));
+                    }
+
+                    const auto end_time{system.CoreTiming().GetClockTicks()};
+
+                    command_buffer.remaining_command_count =
+                        command_list_processor.GetRemainingCommandCount();
+                    command_buffer.render_time_taken = end_time - start_time;
+                }
+            }
+
+            mailbox->ADSPSendMessage(RenderMessage::AudioRenderer_RenderResponse);
+        } break;
+
+        default:
+            LOG_WARNING(Service_Audio,
+                        "ADSP AudioRenderer received an invalid message, msg={:02X}!",
+                        static_cast<u32>(message));
+            break;
+        }
+    }
+}
+
+} // namespace AudioCore::AudioRenderer::ADSP

src/audio_core/renderer/adsp/audio_renderer.h

@@ -0,0 +1,202 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <memory>
+#include <thread>
+
+#include "audio_core/renderer/adsp/command_buffer.h"
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "common/common_types.h"
+#include "common/reader_writer_queue.h"
+#include "common/thread.h"
+
+namespace Core {
+class System;
+}
+
+namespace AudioCore {
+namespace Sink {
+class Sink;
+}
+
+namespace AudioRenderer::ADSP {
+
+enum class RenderMessage {
+    /* 0x00 */ Invalid,
+    /* 0x01 */ AudioRenderer_MapUnmap_Map,
+    /* 0x02 */ AudioRenderer_MapUnmap_MapResponse,
+    /* 0x03 */ AudioRenderer_MapUnmap_Unmap,
+    /* 0x04 */ AudioRenderer_MapUnmap_UnmapResponse,
+    /* 0x05 */ AudioRenderer_MapUnmap_InvalidateCache,
+    /* 0x06 */ AudioRenderer_MapUnmap_InvalidateCacheResponse,
+    /* 0x07 */ AudioRenderer_MapUnmap_Shutdown,
+    /* 0x08 */ AudioRenderer_MapUnmap_ShutdownResponse,
+    /* 0x16 */ AudioRenderer_InitializeOK = 0x16,
+    /* 0x20 */ AudioRenderer_RenderResponse = 0x20,
+    /* 0x2A */ AudioRenderer_Render = 0x2A,
+    /* 0x34 */ AudioRenderer_Shutdown = 0x34,
+};
+
+/**
+ * A mailbox for the AudioRenderer, allowing communication between the host and the AudioRenderer
+ * running on the ADSP.
+ */
+class AudioRenderer_Mailbox {
+public:
+    /**
+     * Send a message from the host to the AudioRenderer.
+     *
+     * @param message_ - The message to send to the AudioRenderer.
+     */
+    void HostSendMessage(RenderMessage message);
+
+    /**
+     * Host wait for a message from the AudioRenderer.
+     *
+     * @return The message returned from the AudioRenderer.
+     */
+    RenderMessage HostWaitMessage();
+
+    /**
+     * Send a message from the AudioRenderer to the host.
+     *
+     * @param message_ - The message to send to the host.
+     */
+    void ADSPSendMessage(RenderMessage message);
+
+    /**
+     * AudioRenderer wait for a message from the host.
+     *
+     * @return The message returned from the AudioRenderer.
+     */
+    RenderMessage ADSPWaitMessage();
+
+    /**
+     * Get the command buffer with the given session id (0 or 1).
+     *
+     * @param session_id - The session id to get (0 or 1).
+     * @return The command buffer.
+     */
+    CommandBuffer& GetCommandBuffer(s32 session_id);
+
+    /**
+     * Set the command buffer with the given session id (0 or 1).
+     *
+     * @param session_id - The session id to get (0 or 1).
+     * @param buffer     - The command buffer to set.
+     */
+    void SetCommandBuffer(u32 session_id, CommandBuffer& buffer);
+
+    /**
+     * Get the total render time taken for the last command lists sent.
+     *
+     * @return Total render time taken for the last command lists.
+     */
+    u64 GetRenderTimeTaken() const;
+
+    /**
+     * Get the tick the AudioRenderer was signalled.
+     *
+     * @return The tick the AudioRenderer was signalled.
+     */
+    u64 GetSignalledTick() const;
+
+    /**
+     * Set the tick the AudioRenderer was signalled.
+     *
+     * @param tick - The tick the AudioRenderer was signalled.
+     */
+    void SetSignalledTick(u64 tick);
+
+    /**
+     * Clear the remaining command count.
+     *
+     * @param session_id - Index for which command list to clear (0 or 1).
+     */
+    void ClearRemainCount(u32 session_id);
+
+    /**
+     * Get the remaining command count for a given command list.
+     *
+     * @param session_id - Index for which command list to clear (0 or 1).
+     * @return The remaining command count.
+     */
+    u32 GetRemainCommandCount(u32 session_id) const;
+
+    /**
+     * Clear the command buffers (does not clear the time taken or the remaining command count).
+     */
+    void ClearCommandBuffers();
+
+private:
+    /// Host signalling event
+    Common::Event host_event{};
+    /// AudioRenderer signalling event
+    Common::Event adsp_event{};
+    /// Host message queue
+
+    Common::ReaderWriterQueue<RenderMessage> host_messages{};
+    /// AudioRenderer message queue
+
+    Common::ReaderWriterQueue<RenderMessage> adsp_messages{};
+    /// Command buffers
+
+    std::array<CommandBuffer, MaxRendererSessions> command_buffers{};
+    /// Tick the AudioRnederer was signalled
+    u64 signalled_tick{};
+};
+
+/**
+ * The AudioRenderer application running on the ADSP.
+ */
+class AudioRenderer {
+public:
+    explicit AudioRenderer(Core::System& system);
+    ~AudioRenderer();
+
+    /**
+     * Start the AudioRenderer.
+     *
+     * @param The mailbox to use for this session.
+     */
+    void Start(AudioRenderer_Mailbox* mailbox);
+
+    /**
+     * Stop the AudioRenderer.
+     */
+    void Stop();
+
+private:
+    /**
+     * Main AudioRenderer thread, responsible for processing the command lists.
+     */
+    void ThreadFunc();
+
+    /**
+     * Creates the streams which will receive the processed samples.
+     */
+    void CreateSinkStreams();
+
+    /// Core system
+    Core::System& system;
+    /// Main thread
+    std::thread thread{};
+    /// The current state
+    std::atomic<bool> running{};
+    /// The active mailbox
+    AudioRenderer_Mailbox* mailbox{};
+    /// The command lists to process
+    std::array<CommandListProcessor, MaxRendererSessions> command_list_processors{};
+    /// The output sink the AudioRenderer will use
+    Sink::Sink& sink;
+    /// The streams which will receive the processed samples
+    std::array<Sink::SinkStream*, MaxRendererSessions> streams;
+    /// An event signalled from the backend when a buffer is consumed, used for timing.
+    Common::Event render_event{};
+};
+
+} // namespace AudioRenderer::ADSP
+} // namespace AudioCore

src/audio_core/renderer/adsp/command_buffer.h

@@ -0,0 +1,21 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "audio_core/common/common.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer::ADSP {
+
+struct CommandBuffer {
+    CpuAddr buffer;
+    u64 size;
+    u64 time_limit;
+    u32 remaining_command_count;
+    bool reset_buffers;
+    u64 applet_resource_user_id;
+    u64 render_time_taken;
+};
+
+} // namespace AudioCore::AudioRenderer::ADSP

src/audio_core/renderer/adsp/command_list_processor.cpp

@@ -0,0 +1,109 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <string>
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/command_list_header.h"
+#include "audio_core/renderer/command/commands.h"
+#include "common/settings.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/core_timing_util.h"
+#include "core/memory.h"
+
+namespace AudioCore::AudioRenderer::ADSP {
+
+void CommandListProcessor::Initialize(Core::System& system_, CpuAddr buffer, u64 size,
+                                      Sink::SinkStream* stream_) {
+    system = &system_;
+    memory = &system->Memory();
+    stream = stream_;
+    header = reinterpret_cast<CommandListHeader*>(buffer);
+    commands = reinterpret_cast<u8*>(buffer + sizeof(CommandListHeader));
+    commands_buffer_size = size;
+    command_count = header->command_count;
+    sample_count = header->sample_count;
+    target_sample_rate = header->sample_rate;
+    mix_buffers = header->samples_buffer;
+    buffer_count = header->buffer_count;
+    processed_command_count = 0;
+}
+
+void CommandListProcessor::SetProcessTimeMax(const u64 time) {
+    max_process_time = time;
+}
+
+u32 CommandListProcessor::GetRemainingCommandCount() const {
+    return command_count - processed_command_count;
+}
+
+void CommandListProcessor::SetBuffer(const CpuAddr buffer, const u64 size) {
+    commands = reinterpret_cast<u8*>(buffer + sizeof(CommandListHeader));
+    commands_buffer_size = size;
+}
+
+Sink::SinkStream* CommandListProcessor::GetOutputSinkStream() const {
+    return stream;
+}
+
+u64 CommandListProcessor::Process(u32 session_id) {
+    const auto start_time_{system->CoreTiming().GetClockTicks()};
+    const auto command_base{CpuAddr(commands)};
+
+    if (processed_command_count > 0) {
+        current_processing_time += start_time_ - end_time;
+    } else {
+        start_time = start_time_;
+        current_processing_time = 0;
+    }
+
+    std::string dump{fmt::format("\nSession {}\n", session_id)};
+
+    for (u32 index = 0; index < command_count; index++) {
+        auto& command{*reinterpret_cast<ICommand*>(commands)};
+
+        if (command.magic != 0xCAFEBABE) {
+            LOG_ERROR(Service_Audio, "Command has invalid magic! Expected 0xCAFEBABE, got {:08X}",
+                      command.magic);
+            return system->CoreTiming().GetClockTicks() - start_time_;
+        }
+
+        auto current_offset{CpuAddr(commands) - command_base};
+
+        if (current_offset + command.size > commands_buffer_size) {
+            LOG_ERROR(Service_Audio,
+                      "Command exceeded command buffer, buffer size {:08X}, command ends at {:08X}",
+                      commands_buffer_size,
+                      CpuAddr(commands) + command.size - sizeof(CommandListHeader));
+            return system->CoreTiming().GetClockTicks() - start_time_;
+        }
+
+        if (Settings::values.dump_audio_commands) {
+            command.Dump(*this, dump);
+        }
+
+        if (!command.Verify(*this)) {
+            break;
+        }
+
+        if (command.enabled) {
+            command.Process(*this);
+        } else {
+            dump += fmt::format("\tDisabled!\n");
+        }
+
+        processed_command_count++;
+        commands += command.size;
+    }
+
+    if (Settings::values.dump_audio_commands && dump != last_dump) {
+        LOG_WARNING(Service_Audio, "{}", dump);
+        last_dump = dump;
+    }
+
+    end_time = system->CoreTiming().GetClockTicks();
+    return end_time - start_time_;
+}
+
+} // namespace AudioCore::AudioRenderer::ADSP

src/audio_core/renderer/adsp/command_list_processor.h

@@ -0,0 +1,118 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <span>
+
+#include "audio_core/common/common.h"
+#include "common/common_types.h"
+
+namespace Core {
+namespace Memory {
+class Memory;
+}
+class System;
+} // namespace Core
+
+namespace AudioCore {
+namespace Sink {
+class SinkStream;
+}
+
+namespace AudioRenderer {
+struct CommandListHeader;
+
+namespace ADSP {
+
+/**
+ * A processor for command lists given to the AudioRenderer.
+ */
+class CommandListProcessor {
+public:
+    /**
+     * Initialize the processor.
+     *
+     * @param system_ - The core system.
+     * @param buffer  - The command buffer to process.
+     * @param size    - The size of the buffer.
+     * @param stream_ - The stream to be used for sending the samples.
+     */
+    void Initialize(Core::System& system, CpuAddr buffer, u64 size, Sink::SinkStream* stream);
+
+    /**
+     * Set the maximum processing time for this command list.
+     *
+     * @param time - The maximum process time.
+     */
+    void SetProcessTimeMax(u64 time);
+
+    /**
+     * Get the remaining command count for this list.
+     *
+     * @return The remaining command count.
+     */
+    u32 GetRemainingCommandCount() const;
+
+    /**
+     * Set the command buffer.
+     *
+     * @param buffer - The buffer to use.
+     * @param size   - The size of the buffer.
+     */
+    void SetBuffer(CpuAddr buffer, u64 size);
+
+    /**
+     * Get the stream for this command list.
+     *
+     * @return The stream associated with this command list.
+     */
+    Sink::SinkStream* GetOutputSinkStream() const;
+
+    /**
+     * Process the command list.
+     *
+     * @param index - Index of the current command list.
+     * @return The time taken to process.
+     */
+    u64 Process(u32 session_id);
+
+    /// Core system
+    Core::System* system{};
+    /// Core memory
+    Core::Memory::Memory* memory{};
+    /// Stream for the processed samples
+    Sink::SinkStream* stream{};
+    /// Header info for this command list
+    CommandListHeader* header{};
+    /// The command buffer
+    u8* commands{};
+    /// The command buffer size
+    u64 commands_buffer_size{};
+    /// The maximum processing time alloted
+    u64 max_process_time{};
+    /// The number of commands in the buffer
+    u32 command_count{};
+    /// The target sample count for output
+    u32 sample_count{};
+    /// The target sample rate for output
+    u32 target_sample_rate{};
+    /// The mixing buffers used by the commands
+    std::span<s32> mix_buffers{};
+    /// The number of mix buffers
+    u32 buffer_count{};
+    /// The number of processed commands so far
+    u32 processed_command_count{};
+    /// The processing start time of this list
+    u64 start_time{};
+    /// The current processing time for this list
+    u64 current_processing_time{};
+    /// The end processing time for this list
+    u64 end_time{};
+    /// Last command list string generated, used for dumping audio commands to console
+    std::string last_dump{};
+};
+
+} // namespace ADSP
+} // namespace AudioRenderer
+} // namespace AudioCore

src/audio_core/renderer/audio_device.cpp

@@ -0,0 +1,52 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_core.h"
+#include "audio_core/common/feature_support.h"
+#include "audio_core/renderer/audio_device.h"
+#include "audio_core/sink/sink.h"
+#include "core/core.h"
+
+namespace AudioCore::AudioRenderer {
+
+AudioDevice::AudioDevice(Core::System& system, const u64 applet_resource_user_id_,
+                         const u32 revision)
+    : output_sink{system.AudioCore().GetOutputSink()},
+      applet_resource_user_id{applet_resource_user_id_}, user_revision{revision} {}
+
+u32 AudioDevice::ListAudioDeviceName(std::vector<AudioDeviceName>& out_buffer,
+                                     const size_t max_count) {
+    std::span<AudioDeviceName> names{};
+
+    if (CheckFeatureSupported(SupportTags::AudioUsbDeviceOutput, user_revision)) {
+        names = usb_device_names;
+    } else {
+        names = device_names;
+    }
+
+    u32 out_count{static_cast<u32>(std::min(max_count, names.size()))};
+    for (u32 i = 0; i < out_count; i++) {
+        out_buffer.push_back(names[i]);
+    }
+    return out_count;
+}
+
+u32 AudioDevice::ListAudioOutputDeviceName(std::vector<AudioDeviceName>& out_buffer,
+                                           const size_t max_count) {
+    u32 out_count{static_cast<u32>(std::min(max_count, output_device_names.size()))};
+
+    for (u32 i = 0; i < out_count; i++) {
+        out_buffer.push_back(output_device_names[i]);
+    }
+    return out_count;
+}
+
+void AudioDevice::SetDeviceVolumes(const f32 volume) {
+    output_sink.SetDeviceVolume(volume);
+}
+
+f32 AudioDevice::GetDeviceVolume([[maybe_unused]] std::string_view name) {
+    return output_sink.GetDeviceVolume();
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/audio_device.h

@@ -0,0 +1,88 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <span>
+
+#include "audio_core/audio_render_manager.h"
+
+namespace Core {
+class System;
+}
+
+namespace AudioCore {
+namespace Sink {
+class Sink;
+}
+
+namespace AudioRenderer {
+/**
+ * An interface to an output audio device available to the Switch.
+ */
+class AudioDevice {
+public:
+    struct AudioDeviceName {
+        std::array<char, 0x100> name;
+
+        AudioDeviceName(const char* name_) {
+            std::strncpy(name.data(), name_, name.size());
+        }
+    };
+
+    std::array<AudioDeviceName, 4> usb_device_names{"AudioStereoJackOutput",
+                                                    "AudioBuiltInSpeakerOutput", "AudioTvOutput",
+                                                    "AudioUsbDeviceOutput"};
+    std::array<AudioDeviceName, 3> device_names{"AudioStereoJackOutput",
+                                                "AudioBuiltInSpeakerOutput", "AudioTvOutput"};
+    std::array<AudioDeviceName, 3> output_device_names{"AudioBuiltInSpeakerOutput", "AudioTvOutput",
+                                                       "AudioExternalOutput"};
+
+    explicit AudioDevice(Core::System& system, u64 applet_resource_user_id, u32 revision);
+
+    /**
+     * Get a list of the available output devices.
+     *
+     * @param out_buffer - Output buffer to write the available device names.
+     * @param max_count  - Maximum number of devices to write (count of out_buffer).
+     * @return Number of device names written.
+     */
+    u32 ListAudioDeviceName(std::vector<AudioDeviceName>& out_buffer, size_t max_count);
+
+    /**
+     * Get a list of the available output devices.
+     * Different to above somehow...
+     *
+     * @param out_buffer - Output buffer to write the available device names.
+     * @param max_count  - Maximum number of devices to write (count of out_buffer).
+     * @return Number of device names written.
+     */
+    u32 ListAudioOutputDeviceName(std::vector<AudioDeviceName>& out_buffer, size_t max_count);
+
+    /**
+     * Set the volume of all streams in the backend sink.
+     *
+     * @param volume - Volume to set.
+     */
+    void SetDeviceVolumes(f32 volume);
+
+    /**
+     * Get the volume for a given device name.
+     * Note: This is not fully implemented, we only assume 1 device for all streams.
+     *
+     * @param name - Name of the device to check. Unused.
+     * @return Volume of the device.
+     */
+    f32 GetDeviceVolume(std::string_view name);
+
+private:
+    /// Backend output sink for the device
+    Sink::Sink& output_sink;
+    /// Resource id this device is used for
+    const u64 applet_resource_user_id;
+    /// User audio renderer revision
+    const u32 user_revision;
+};
+
+} // namespace AudioRenderer
+} // namespace AudioCore

src/audio_core/renderer/audio_renderer.cpp

@@ -0,0 +1,67 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_render_manager.h"
+#include "audio_core/common/audio_renderer_parameter.h"
+#include "audio_core/renderer/audio_renderer.h"
+#include "audio_core/renderer/system_manager.h"
+#include "core/core.h"
+#include "core/hle/kernel/k_transfer_memory.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace AudioCore::AudioRenderer {
+
+Renderer::Renderer(Core::System& system_, Manager& manager_, Kernel::KEvent* rendered_event)
+    : core{system_}, manager{manager_}, system{system_, rendered_event} {}
+
+Result Renderer::Initialize(const AudioRendererParameterInternal& params,
+                            Kernel::KTransferMemory* transfer_memory,
+                            const u64 transfer_memory_size, const u32 process_handle,
+                            const u64 applet_resource_user_id, const s32 session_id) {
+    if (params.execution_mode == ExecutionMode::Auto) {
+        if (!manager.AddSystem(system)) {
+            LOG_ERROR(Service_Audio,
+                      "Both Audio Render sessions are in use, cannot create any more");
+            return Service::Audio::ERR_MAXIMUM_SESSIONS_REACHED;
+        }
+        system_registered = true;
+    }
+
+    initialized = true;
+    system.Initialize(params, transfer_memory, transfer_memory_size, process_handle,
+                      applet_resource_user_id, session_id);
+
+    return ResultSuccess;
+}
+
+void Renderer::Finalize() {
+    auto session_id{system.GetSessionId()};
+
+    system.Finalize();
+
+    if (system_registered) {
+        manager.RemoveSystem(system);
+        system_registered = false;
+    }
+
+    manager.ReleaseSessionId(session_id);
+}
+
+System& Renderer::GetSystem() {
+    return system;
+}
+
+void Renderer::Start() {
+    system.Start();
+}
+
+void Renderer::Stop() {
+    system.Stop();
+}
+
+Result Renderer::RequestUpdate(std::span<const u8> input, std::span<u8> performance,
+                               std::span<u8> output) {
+    return system.Update(input, performance, output);
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/audio_renderer.h

@@ -0,0 +1,97 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <span>
+
+#include "audio_core/renderer/system.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace Core {
+class System;
+}
+
+namespace Kernel {
+class KTransferMemory;
+}
+
+namespace AudioCore {
+struct AudioRendererParameterInternal;
+
+namespace AudioRenderer {
+class Manager;
+
+/**
+ * Audio Renderer, wraps the main audio system and is mainly responsible for handling service calls.
+ */
+class Renderer {
+public:
+    explicit Renderer(Core::System& system, Manager& manager, Kernel::KEvent* rendered_event);
+
+    /**
+     * Initialize the renderer.
+     * Registers the system with the AudioRenderer::Manager, allocates workbuffers and initializes
+     * everything to a default state.
+     *
+     * @param params                  - Input parameters to initialize the system with.
+     * @param transfer_memory         - Game-supplied memory for all workbuffers. Unused.
+     * @param transfer_memory_size    - Size of the transfer memory. Unused.
+     * @param process_handle          - Process handle, also used for memory. Unused.
+     * @param applet_resource_user_id - Applet id for this renderer. Unused.
+     * @param session_id              - Session id of this renderer.
+     * @return Result code.
+     */
+    Result Initialize(const AudioRendererParameterInternal& params,
+                      Kernel::KTransferMemory* transfer_memory, u64 transfer_memory_size,
+                      u32 process_handle, u64 applet_resource_user_id, s32 session_id);
+
+    /**
+     * Finalize the renderer for shutdown.
+     */
+    void Finalize();
+
+    /**
+     * Get the renderer's system.
+     *
+     * @return Reference to the system.
+     */
+    System& GetSystem();
+
+    /**
+     * Start the renderer.
+     */
+    void Start();
+
+    /**
+     * Stop the renderer.
+     */
+    void Stop();
+
+    /**
+     * Update the audio renderer with new information.
+     * Called via RequestUpdate from the AudRen:U service.
+     *
+     * @param input       - Input buffer containing the new data.
+     * @param performance - Optional performance buffer for outputting performance metrics.
+     * @param output      - Output data from the renderer.
+     * @return Result code.
+     */
+    Result RequestUpdate(std::span<const u8> input, std::span<u8> performance,
+                         std::span<u8> output);
+
+private:
+    /// System core
+    Core::System& core;
+    /// Manager this renderer is registered with
+    Manager& manager;
+    /// Is the audio renderer initialized?
+    bool initialized{};
+    /// Is the system registered with the manager?
+    bool system_registered{};
+    /// Audio render system, main driver of audio rendering
+    System system;
+};
+
+} // namespace AudioRenderer
+} // namespace AudioCore

src/audio_core/renderer/behavior/behavior_info.cpp

@@ -0,0 +1,174 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/common/feature_support.h"
+#include "audio_core/renderer/behavior/behavior_info.h"
+
+namespace AudioCore::AudioRenderer {
+
+BehaviorInfo::BehaviorInfo() : process_revision{CurrentRevision} {}
+
+u32 BehaviorInfo::GetProcessRevisionNum() const {
+    return process_revision;
+}
+
+u32 BehaviorInfo::GetProcessRevision() const {
+    return Common::MakeMagic('R', 'E', 'V',
+                             static_cast<char>(static_cast<u8>('0') + process_revision));
+}
+
+u32 BehaviorInfo::GetUserRevisionNum() const {
+    return user_revision;
+}
+
+u32 BehaviorInfo::GetUserRevision() const {
+    return Common::MakeMagic('R', 'E', 'V',
+                             static_cast<char>(static_cast<u8>('0') + user_revision));
+}
+
+void BehaviorInfo::SetUserLibRevision(const u32 user_revision_) {
+    user_revision = GetRevisionNum(user_revision_);
+}
+
+void BehaviorInfo::ClearError() {
+    error_count = 0;
+}
+
+void BehaviorInfo::AppendError(ErrorInfo& error) {
+    LOG_ERROR(Service_Audio, "Error during RequestUpdate, reporting code {:04X} address {:08X}",
+              error.error_code.raw, error.address);
+    if (error_count < MaxErrors) {
+        errors[error_count++] = error;
+    }
+}
+
+void BehaviorInfo::CopyErrorInfo(std::span<ErrorInfo> out_errors, u32& out_count) {
+    auto error_count_{std::min(error_count, MaxErrors)};
+    std::memset(out_errors.data(), 0, MaxErrors * sizeof(ErrorInfo));
+
+    for (size_t i = 0; i < error_count_; i++) {
+        out_errors[i] = errors[i];
+    }
+    out_count = error_count_;
+}
+
+void BehaviorInfo::UpdateFlags(const Flags flags_) {
+    flags = flags_;
+}
+
+bool BehaviorInfo::IsMemoryForceMappingEnabled() const {
+    return flags.IsMemoryForceMappingEnabled;
+}
+
+bool BehaviorInfo::IsAdpcmLoopContextBugFixed() const {
+    return CheckFeatureSupported(SupportTags::AdpcmLoopContextBugFix, user_revision);
+}
+
+bool BehaviorInfo::IsSplitterSupported() const {
+    return CheckFeatureSupported(SupportTags::Splitter, user_revision);
+}
+
+bool BehaviorInfo::IsSplitterBugFixed() const {
+    return CheckFeatureSupported(SupportTags::SplitterBugFix, user_revision);
+}
+
+bool BehaviorInfo::IsEffectInfoVersion2Supported() const {
+    return CheckFeatureSupported(SupportTags::EffectInfoVer2, user_revision);
+}
+
+bool BehaviorInfo::IsVariadicCommandBufferSizeSupported() const {
+    return CheckFeatureSupported(SupportTags::AudioRendererVariadicCommandBufferSize,
+                                 user_revision);
+}
+
+bool BehaviorInfo::IsWaveBufferVer2Supported() const {
+    return CheckFeatureSupported(SupportTags::WaveBufferVer2, user_revision);
+}
+
+bool BehaviorInfo::IsLongSizePreDelaySupported() const {
+    return CheckFeatureSupported(SupportTags::LongSizePreDelay, user_revision);
+}
+
+bool BehaviorInfo::IsCommandProcessingTimeEstimatorVersion2Supported() const {
+    return CheckFeatureSupported(SupportTags::CommandProcessingTimeEstimatorVersion2,
+                                 user_revision);
+}
+
+bool BehaviorInfo::IsCommandProcessingTimeEstimatorVersion3Supported() const {
+    return CheckFeatureSupported(SupportTags::CommandProcessingTimeEstimatorVersion3,
+                                 user_revision);
+}
+
+bool BehaviorInfo::IsCommandProcessingTimeEstimatorVersion4Supported() const {
+    return CheckFeatureSupported(SupportTags::CommandProcessingTimeEstimatorVersion4,
+                                 user_revision);
+}
+
+bool BehaviorInfo::IsAudioRenererProcessingTimeLimit70PercentSupported() const {
+    return CheckFeatureSupported(SupportTags::AudioRendererProcessingTimeLimit70Percent,
+                                 user_revision);
+}
+
+bool BehaviorInfo::IsAudioRenererProcessingTimeLimit75PercentSupported() const {
+    return CheckFeatureSupported(SupportTags::AudioRendererProcessingTimeLimit75Percent,
+                                 user_revision);
+}
+
+bool BehaviorInfo::IsAudioRenererProcessingTimeLimit80PercentSupported() const {
+    return CheckFeatureSupported(SupportTags::AudioRendererProcessingTimeLimit80Percent,
+                                 user_revision);
+}
+
+bool BehaviorInfo::IsFlushVoiceWaveBuffersSupported() const {
+    return CheckFeatureSupported(SupportTags::FlushVoiceWaveBuffers, user_revision);
+}
+
+bool BehaviorInfo::IsElapsedFrameCountSupported() const {
+    return CheckFeatureSupported(SupportTags::ElapsedFrameCount, user_revision);
+}
+
+bool BehaviorInfo::IsPerformanceMetricsDataFormatVersion2Supported() const {
+    return CheckFeatureSupported(SupportTags::PerformanceMetricsDataFormatVersion2, user_revision);
+}
+
+size_t BehaviorInfo::GetPerformanceMetricsDataFormat() const {
+    if (CheckFeatureSupported(SupportTags::PerformanceMetricsDataFormatVersion2, user_revision)) {
+        return 2;
+    }
+    return 1;
+}
+
+bool BehaviorInfo::IsVoicePitchAndSrcSkippedSupported() const {
+    return CheckFeatureSupported(SupportTags::VoicePitchAndSrcSkipped, user_revision);
+}
+
+bool BehaviorInfo::IsVoicePlayedSampleCountResetAtLoopPointSupported() const {
+    return CheckFeatureSupported(SupportTags::VoicePlayedSampleCountResetAtLoopPoint,
+                                 user_revision);
+}
+
+bool BehaviorInfo::IsBiquadFilterEffectStateClaerBugFixed() const {
+    return CheckFeatureSupported(SupportTags::BiquadFilterEffectStateClearBugFix, user_revision);
+}
+
+bool BehaviorInfo::IsVolumeMixParameterPrecisionQ23Supported() const {
+    return CheckFeatureSupported(SupportTags::VolumeMixParameterPrecisionQ23, user_revision);
+}
+
+bool BehaviorInfo::UseBiquadFilterFloatProcessing() const {
+    return CheckFeatureSupported(SupportTags::BiquadFilterFloatProcessing, user_revision);
+}
+
+bool BehaviorInfo::IsMixInParameterDirtyOnlyUpdateSupported() const {
+    return CheckFeatureSupported(SupportTags::MixInParameterDirtyOnlyUpdate, user_revision);
+}
+
+bool BehaviorInfo::UseMultiTapBiquadFilterProcessing() const {
+    return CheckFeatureSupported(SupportTags::MultiTapBiquadFilterProcessing, user_revision);
+}
+
+bool BehaviorInfo::IsDeviceApiVersion2Supported() const {
+    return CheckFeatureSupported(SupportTags::DeviceApiVersion2, user_revision);
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/behavior/behavior_info.h

@@ -0,0 +1,337 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <span>
+
+#include "audio_core/common/common.h"
+#include "common/common_types.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace AudioCore::AudioRenderer {
+/**
+ * Holds host and user revisions, checks whether render features can be enabled, and reports errors.
+ */
+class BehaviorInfo {
+    static constexpr u32 MaxErrors = 10;
+
+public:
+    struct ErrorInfo {
+        /* 0x00 */ Result error_code{0};
+        /* 0x04 */ u32 unk_04;
+        /* 0x08 */ CpuAddr address;
+    };
+    static_assert(sizeof(ErrorInfo) == 0x10, "BehaviorInfo::ErrorInfo has the wrong size!");
+
+    struct Flags {
+        u64 IsMemoryForceMappingEnabled : 1;
+    };
+
+    struct InParameter {
+        /* 0x00 */ u32 revision;
+        /* 0x08 */ Flags flags;
+    };
+    static_assert(sizeof(InParameter) == 0x10, "BehaviorInfo::InParameter has the wrong size!");
+
+    struct OutStatus {
+        /* 0x00 */ std::array<ErrorInfo, MaxErrors> errors;
+        /* 0xA0 */ u32 error_count;
+        /* 0xA4 */ char unkA4[0xC];
+    };
+    static_assert(sizeof(OutStatus) == 0xB0, "BehaviorInfo::OutStatus has the wrong size!");
+
+    BehaviorInfo();
+
+    /**
+     * Get the host revision as a number.
+     *
+     * @return The host revision.
+     */
+    u32 GetProcessRevisionNum() const;
+
+    /**
+     * Get the host revision in chars, e.g REV8.
+     * Rev 10 and higher use the ascii characters above 9.
+     * E.g:
+     *     Rev 10 = REV:
+     *     Rev 11 = REV;
+     *
+     * @return The host revision.
+     */
+    u32 GetProcessRevision() const;
+
+    /**
+     * Get the user revision as a number.
+     *
+     * @return The user revision.
+     */
+    u32 GetUserRevisionNum() const;
+
+    /**
+     * Get the user revision in chars, e.g REV8.
+     * Rev 10 and higher use the ascii characters above 9. REV: REV; etc.
+     *
+     * @return The user revision.
+     */
+    u32 GetUserRevision() const;
+
+    /**
+     * Set the user revision.
+     *
+     * @param user_revision - The user's revision.
+     */
+    void SetUserLibRevision(u32 user_revision);
+
+    /**
+     * Clear the current error count.
+     */
+    void ClearError();
+
+    /**
+     * Append an error to the error list.
+     *
+     * @param error - The new error.
+     */
+    void AppendError(ErrorInfo& error);
+
+    /**
+     * Copy errors to the given output container.
+     *
+     * @param out_errors - Output container to receive the errors.
+     * @param out_count  - The number of errors written.
+     */
+    void CopyErrorInfo(std::span<ErrorInfo> out_errors, u32& out_count);
+
+    /**
+     * Update the behaviour flags.
+     *
+     * @param flags - New flags to use.
+     */
+    void UpdateFlags(Flags flags);
+
+    /**
+     * Check if memory pools can be forcibly mapped.
+     *
+     * @return True if enabled, otherwise false.
+     */
+    bool IsMemoryForceMappingEnabled() const;
+
+    /**
+     * Check if the ADPCM context bug is fixed.
+     * The ADPCM context was not being sent to the AudioRenderer, leading to incorrect scaling being
+     * used.
+     *
+     * @return True if fixed, otherwise false.
+     */
+    bool IsAdpcmLoopContextBugFixed() const;
+
+    /**
+     * Check if the splitter is supported.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsSplitterSupported() const;
+
+    /**
+     * Check if the splitter bug is fixed.
+     * Update is given the wrong number of splitter destinations, leading to invalid data
+     * being processed.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsSplitterBugFixed() const;
+
+    /**
+     * Check if effects version 2 are supported.
+     * This gives support for returning effect states from the AudioRenderer, currently only used
+     * for Limiter statistics.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsEffectInfoVersion2Supported() const;
+
+    /**
+     * Check if a variadic command buffer is supported.
+     * As of Rev 5 with the added optional performance metric logging, the command
+     * buffer can be a variable size, so take that into account for calcualting its size.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsVariadicCommandBufferSizeSupported() const;
+
+    /**
+     * Check if wave buffers version 2 are supported.
+     * See WaveBufferVersion1 and WaveBufferVersion2.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsWaveBufferVer2Supported() const;
+
+    /**
+     * Check if long size pre delay is supported.
+     * This allows a longer initial delay time for the Reverb command.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsLongSizePreDelaySupported() const;
+
+    /**
+     * Check if the command time estimator version 2 is supported.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsCommandProcessingTimeEstimatorVersion2Supported() const;
+
+    /**
+     * Check if the command time estimator version 3 is supported.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsCommandProcessingTimeEstimatorVersion3Supported() const;
+
+    /**
+     * Check if the command time estimator version 4 is supported.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsCommandProcessingTimeEstimatorVersion4Supported() const;
+
+    /**
+     * Check if the AudioRenderer can use up to 70% of the allocated processing timeslice.
+     * Note: Name is correct, Nintendo have the typo here
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsAudioRenererProcessingTimeLimit70PercentSupported() const;
+
+    /**
+     * Check if the AudioRenderer can use up to 75% of the allocated processing timeslice.
+     * Note: Name is correct, Nintendo have the typo here
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsAudioRenererProcessingTimeLimit75PercentSupported() const;
+
+    /**
+     * Check if the AudioRenderer can use up to 80% of the allocated processing timeslice.
+     * Note: Name is correct, Nintendo have the typo here
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsAudioRenererProcessingTimeLimit80PercentSupported() const;
+
+    /**
+     * Check if voice flushing is supported
+     * This allowws low-priority voices to be dropped if the AudioRenderer is running behind.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsFlushVoiceWaveBuffersSupported() const;
+
+    /**
+     * Check if counting the number of elapsed frames is supported.
+     * This adds extra output to RequestUpdate, returning the number of times the AudioRenderer
+     * processed a command list.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsElapsedFrameCountSupported() const;
+
+    /**
+     * Check if performance metrics version 2 are supported.
+     * This adds extra output to RequestUpdate, returning the number of times the AudioRenderer
+     * (Unused?).
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsPerformanceMetricsDataFormatVersion2Supported() const;
+
+    /**
+     * Get the supported performance metrics version.
+     * Version 2 logs some extra fields in output, such as number of voices dropped,
+     * processing start time, if the AudioRenderer exceeded its time, etc.
+     *
+     * @return Version supported, either 1 or 2.
+     */
+    size_t GetPerformanceMetricsDataFormat() const;
+
+    /**
+     * Check if skipping voice pitch and sample rate conversion is supported.
+     * This speeds up the data source commands by skipping resampling if unwanted.
+     * See AudioCore::AudioRenderer::DecodeFromWaveBuffers
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsVoicePitchAndSrcSkippedSupported() const;
+
+    /**
+     * Check if resetting played sample count at loop points is supported.
+     * This resets the number of samples played in a voice state when a loop point is reached.
+     * See AudioCore::AudioRenderer::DecodeFromWaveBuffers
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsVoicePlayedSampleCountResetAtLoopPointSupported() const;
+
+    /**
+     * Check if the clear state bug for biquad filters is fixed.
+     * The biquad state was not marked as needing re-initialisation when the effect was updated, it
+     * was only initialized once with a new effect.
+     * Note: Name is correct, Nintendo have the typo here
+     *
+     * @return True if fixed, otherwise false.
+     */
+    bool IsBiquadFilterEffectStateClaerBugFixed() const;
+
+    /**
+     * Check if Q23 precision is supported for fixed point.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsVolumeMixParameterPrecisionQ23Supported() const;
+
+    /**
+     * Check if float processing for biuad filters is supported.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool UseBiquadFilterFloatProcessing() const;
+
+    /**
+     * Check if dirty-only mix updates are supported.
+     * This saves a lot of buffer size as mixes can be large and not change much.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsMixInParameterDirtyOnlyUpdateSupported() const;
+
+    /**
+     * Check if multi-tap biquad filters are supported.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool UseMultiTapBiquadFilterProcessing() const;
+
+    /**
+     * Check if device api version 2 is supported.
+     * In the SDK but not in any sysmodule? Not sure, left here for completeness anyway.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsDeviceApiVersion2Supported() const;
+
+    /// Host version
+    u32 process_revision;
+    /// User version
+    u32 user_revision{};
+    /// Behaviour flags
+    Flags flags{};
+    /// Errors generated and reported during Update
+    std::array<ErrorInfo, MaxErrors> errors{};
+    /// Error count
+    u32 error_count{};
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/behavior/info_updater.cpp

@@ -0,0 +1,539 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/common/feature_support.h"
+#include "audio_core/renderer/behavior/behavior_info.h"
+#include "audio_core/renderer/behavior/info_updater.h"
+#include "audio_core/renderer/effect/effect_context.h"
+#include "audio_core/renderer/effect/effect_reset.h"
+#include "audio_core/renderer/memory/memory_pool_info.h"
+#include "audio_core/renderer/mix/mix_context.h"
+#include "audio_core/renderer/performance/performance_manager.h"
+#include "audio_core/renderer/sink/circular_buffer_sink_info.h"
+#include "audio_core/renderer/sink/device_sink_info.h"
+#include "audio_core/renderer/sink/sink_context.h"
+#include "audio_core/renderer/splitter/splitter_context.h"
+#include "audio_core/renderer/voice/voice_context.h"
+
+namespace AudioCore::AudioRenderer {
+
+InfoUpdater::InfoUpdater(std::span<const u8> input_, std::span<u8> output_,
+                         const u32 process_handle_, BehaviorInfo& behaviour_)
+    : input{input_.data() + sizeof(UpdateDataHeader)},
+      input_origin{input_}, output{output_.data() + sizeof(UpdateDataHeader)},
+      output_origin{output_}, in_header{reinterpret_cast<const UpdateDataHeader*>(
+                                  input_origin.data())},
+      out_header{reinterpret_cast<UpdateDataHeader*>(output_origin.data())},
+      expected_input_size{input_.size()}, expected_output_size{output_.size()},
+      process_handle{process_handle_}, behaviour{behaviour_} {
+    std::construct_at<UpdateDataHeader>(out_header, behaviour.GetProcessRevision());
+}
+
+Result InfoUpdater::UpdateVoiceChannelResources(VoiceContext& voice_context) {
+    const auto voice_count{voice_context.GetCount()};
+    std::span<const VoiceChannelResource::InParameter> in_params{
+        reinterpret_cast<const VoiceChannelResource::InParameter*>(input), voice_count};
+
+    for (u32 i = 0; i < voice_count; i++) {
+        auto& resource{voice_context.GetChannelResource(i)};
+        resource.in_use = in_params[i].in_use;
+        if (in_params[i].in_use) {
+            resource.mix_volumes = in_params[i].mix_volumes;
+        }
+    }
+
+    const auto consumed_input_size{voice_count *
+                                   static_cast<u32>(sizeof(VoiceChannelResource::InParameter))};
+    if (consumed_input_size != in_header->voice_resources_size) {
+        LOG_ERROR(Service_Audio,
+                  "Consumed an incorrect voice resource size, header size={}, consumed={}",
+                  in_header->voice_resources_size, consumed_input_size);
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    input += consumed_input_size;
+    return ResultSuccess;
+}
+
+Result InfoUpdater::UpdateVoices(VoiceContext& voice_context,
+                                 std::span<MemoryPoolInfo> memory_pools,
+                                 const u32 memory_pool_count) {
+    const PoolMapper pool_mapper(process_handle, memory_pools, memory_pool_count,
+                                 behaviour.IsMemoryForceMappingEnabled());
+    const auto voice_count{voice_context.GetCount()};
+    std::span<const VoiceInfo::InParameter> in_params{
+        reinterpret_cast<const VoiceInfo::InParameter*>(input), voice_count};
+    std::span<VoiceInfo::OutStatus> out_params{reinterpret_cast<VoiceInfo::OutStatus*>(output),
+                                               voice_count};
+
+    for (u32 i = 0; i < voice_count; i++) {
+        auto& voice_info{voice_context.GetInfo(i)};
+        voice_info.in_use = false;
+    }
+
+    u32 new_voice_count{0};
+
+    for (u32 i = 0; i < voice_count; i++) {
+        const auto& in_param{in_params[i]};
+        std::array<VoiceState*, MaxChannels> voice_states{};
+
+        if (!in_param.in_use) {
+            continue;
+        }
+
+        auto& voice_info{voice_context.GetInfo(in_param.id)};
+
+        for (u32 channel = 0; channel < in_param.channel_count; channel++) {
+            voice_states[channel] = &voice_context.GetState(in_param.channel_resource_ids[channel]);
+        }
+
+        if (in_param.is_new) {
+            voice_info.Initialize();
+
+            for (u32 channel = 0; channel < in_param.channel_count; channel++) {
+                std::memset(voice_states[channel], 0, sizeof(VoiceState));
+            }
+        }
+
+        BehaviorInfo::ErrorInfo update_error{};
+        voice_info.UpdateParameters(update_error, in_param, pool_mapper, behaviour);
+
+        if (!update_error.error_code.IsSuccess()) {
+            behaviour.AppendError(update_error);
+        }
+
+        std::array<std::array<BehaviorInfo::ErrorInfo, 2>, MaxWaveBuffers> wavebuffer_errors{};
+        voice_info.UpdateWaveBuffers(wavebuffer_errors, MaxWaveBuffers * 2, in_param, voice_states,
+                                     pool_mapper, behaviour);
+
+        for (auto& wavebuffer_error : wavebuffer_errors) {
+            for (auto& error : wavebuffer_error) {
+                if (error.error_code.IsError()) {
+                    behaviour.AppendError(error);
+                }
+            }
+        }
+
+        voice_info.WriteOutStatus(out_params[i], in_param, voice_states);
+        new_voice_count += in_param.channel_count;
+    }
+
+    auto consumed_input_size{voice_count * static_cast<u32>(sizeof(VoiceInfo::InParameter))};
+    auto consumed_output_size{voice_count * static_cast<u32>(sizeof(VoiceInfo::OutStatus))};
+    if (consumed_input_size != in_header->voices_size) {
+        LOG_ERROR(Service_Audio, "Consumed an incorrect voices size, header size={}, consumed={}",
+                  in_header->voices_size, consumed_input_size);
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    out_header->voices_size = consumed_output_size;
+    out_header->size += consumed_output_size;
+    input += consumed_input_size;
+    output += consumed_output_size;
+
+    voice_context.SetActiveCount(new_voice_count);
+
+    return ResultSuccess;
+}
+
+Result InfoUpdater::UpdateEffects(EffectContext& effect_context, const bool renderer_active,
+                                  std::span<MemoryPoolInfo> memory_pools,
+                                  const u32 memory_pool_count) {
+    if (behaviour.IsEffectInfoVersion2Supported()) {
+        return UpdateEffectsVersion2(effect_context, renderer_active, memory_pools,
+                                     memory_pool_count);
+    } else {
+        return UpdateEffectsVersion1(effect_context, renderer_active, memory_pools,
+                                     memory_pool_count);
+    }
+}
+
+Result InfoUpdater::UpdateEffectsVersion1(EffectContext& effect_context, const bool renderer_active,
+                                          std::span<MemoryPoolInfo> memory_pools,
+                                          const u32 memory_pool_count) {
+    PoolMapper pool_mapper(process_handle, memory_pools, memory_pool_count,
+                           behaviour.IsMemoryForceMappingEnabled());
+
+    const auto effect_count{effect_context.GetCount()};
+
+    std::span<const EffectInfoBase::InParameterVersion1> in_params{
+        reinterpret_cast<const EffectInfoBase::InParameterVersion1*>(input), effect_count};
+    std::span<EffectInfoBase::OutStatusVersion1> out_params{
+        reinterpret_cast<EffectInfoBase::OutStatusVersion1*>(output), effect_count};
+
+    for (u32 i = 0; i < effect_count; i++) {
+        auto effect_info{&effect_context.GetInfo(i)};
+        if (effect_info->GetType() != in_params[i].type) {
+            effect_info->ForceUnmapBuffers(pool_mapper);
+            ResetEffect(effect_info, in_params[i].type);
+        }
+
+        BehaviorInfo::ErrorInfo error_info{};
+        effect_info->Update(error_info, in_params[i], pool_mapper);
+        if (error_info.error_code.IsError()) {
+            behaviour.AppendError(error_info);
+        }
+
+        effect_info->StoreStatus(out_params[i], renderer_active);
+    }
+
+    auto consumed_input_size{effect_count *
+                             static_cast<u32>(sizeof(EffectInfoBase::InParameterVersion1))};
+    auto consumed_output_size{effect_count *
+                              static_cast<u32>(sizeof(EffectInfoBase::OutStatusVersion1))};
+    if (consumed_input_size != in_header->effects_size) {
+        LOG_ERROR(Service_Audio, "Consumed an incorrect effects size, header size={}, consumed={}",
+                  in_header->effects_size, consumed_input_size);
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    out_header->effects_size = consumed_output_size;
+    out_header->size += consumed_output_size;
+    input += consumed_input_size;
+    output += consumed_output_size;
+
+    return ResultSuccess;
+}
+
+Result InfoUpdater::UpdateEffectsVersion2(EffectContext& effect_context, const bool renderer_active,
+                                          std::span<MemoryPoolInfo> memory_pools,
+                                          const u32 memory_pool_count) {
+    PoolMapper pool_mapper(process_handle, memory_pools, memory_pool_count,
+                           behaviour.IsMemoryForceMappingEnabled());
+
+    const auto effect_count{effect_context.GetCount()};
+
+    std::span<const EffectInfoBase::InParameterVersion2> in_params{
+        reinterpret_cast<const EffectInfoBase::InParameterVersion2*>(input), effect_count};
+    std::span<EffectInfoBase::OutStatusVersion2> out_params{
+        reinterpret_cast<EffectInfoBase::OutStatusVersion2*>(output), effect_count};
+
+    for (u32 i = 0; i < effect_count; i++) {
+        auto effect_info{&effect_context.GetInfo(i)};
+        if (effect_info->GetType() != in_params[i].type) {
+            effect_info->ForceUnmapBuffers(pool_mapper);
+            ResetEffect(effect_info, in_params[i].type);
+        }
+
+        BehaviorInfo::ErrorInfo error_info{};
+        effect_info->Update(error_info, in_params[i], pool_mapper);
+
+        if (error_info.error_code.IsError()) {
+            behaviour.AppendError(error_info);
+        }
+
+        effect_info->StoreStatus(out_params[i], renderer_active);
+
+        if (in_params[i].is_new) {
+            effect_info->InitializeResultState(effect_context.GetDspSharedResultState(i));
+            effect_info->InitializeResultState(effect_context.GetResultState(i));
+        }
+        effect_info->UpdateResultState(out_params[i].result_state,
+                                       effect_context.GetResultState(i));
+    }
+
+    auto consumed_input_size{effect_count *
+                             static_cast<u32>(sizeof(EffectInfoBase::InParameterVersion2))};
+    auto consumed_output_size{effect_count *
+                              static_cast<u32>(sizeof(EffectInfoBase::OutStatusVersion2))};
+    if (consumed_input_size != in_header->effects_size) {
+        LOG_ERROR(Service_Audio, "Consumed an incorrect effects size, header size={}, consumed={}",
+                  in_header->effects_size, consumed_input_size);
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    out_header->effects_size = consumed_output_size;
+    out_header->size += consumed_output_size;
+    input += consumed_input_size;
+    output += consumed_output_size;
+
+    return ResultSuccess;
+}
+
+Result InfoUpdater::UpdateMixes(MixContext& mix_context, const u32 mix_buffer_count,
+                                EffectContext& effect_context, SplitterContext& splitter_context) {
+    s32 mix_count{0};
+    u32 consumed_input_size{0};
+
+    if (behaviour.IsMixInParameterDirtyOnlyUpdateSupported()) {
+        auto in_dirty_params{reinterpret_cast<const MixInfo::InDirtyParameter*>(input)};
+        mix_count = in_dirty_params->count;
+        input += sizeof(MixInfo::InDirtyParameter);
+        consumed_input_size = static_cast<u32>(sizeof(MixInfo::InDirtyParameter) +
+                                               mix_count * sizeof(MixInfo::InParameter));
+    } else {
+        mix_count = mix_context.GetCount();
+        consumed_input_size = static_cast<u32>(mix_count * sizeof(MixInfo::InParameter));
+    }
+
+    if (mix_buffer_count == 0) {
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    std::span<const MixInfo::InParameter> in_params{
+        reinterpret_cast<const MixInfo::InParameter*>(input), static_cast<size_t>(mix_count)};
+
+    u32 total_buffer_count{0};
+    for (s32 i = 0; i < mix_count; i++) {
+        const auto& params{in_params[i]};
+
+        if (params.in_use) {
+            total_buffer_count += params.buffer_count;
+            if (params.dest_mix_id > static_cast<s32>(mix_context.GetCount()) &&
+                params.dest_mix_id != UnusedMixId && params.mix_id != FinalMixId) {
+                return Service::Audio::ERR_INVALID_UPDATE_DATA;
+            }
+        }
+    }
+
+    if (total_buffer_count > mix_buffer_count) {
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    bool mix_dirty{false};
+    for (s32 i = 0; i < mix_count; i++) {
+        const auto& params{in_params[i]};
+
+        s32 mix_id{i};
+        if (behaviour.IsMixInParameterDirtyOnlyUpdateSupported()) {
+            mix_id = params.mix_id;
+        }
+
+        auto mix_info{mix_context.GetInfo(mix_id)};
+        if (mix_info->in_use != params.in_use) {
+            mix_info->in_use = params.in_use;
+            if (!params.in_use) {
+                mix_info->ClearEffectProcessingOrder();
+            }
+            mix_dirty = true;
+        }
+
+        if (params.in_use) {
+            mix_dirty |= mix_info->Update(mix_context.GetEdgeMatrix(), params, effect_context,
+                                          splitter_context, behaviour);
+        }
+    }
+
+    if (mix_dirty) {
+        if (behaviour.IsSplitterSupported() && splitter_context.UsingSplitter()) {
+            if (!mix_context.TSortInfo(splitter_context)) {
+                return Service::Audio::ERR_INVALID_UPDATE_DATA;
+            }
+        } else {
+            mix_context.SortInfo();
+        }
+    }
+
+    if (consumed_input_size != in_header->mix_size) {
+        LOG_ERROR(Service_Audio, "Consumed an incorrect mixes size, header size={}, consumed={}",
+                  in_header->mix_size, consumed_input_size);
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    input += mix_count * sizeof(MixInfo::InParameter);
+
+    return ResultSuccess;
+}
+
+Result InfoUpdater::UpdateSinks(SinkContext& sink_context, std::span<MemoryPoolInfo> memory_pools,
+                                const u32 memory_pool_count) {
+    PoolMapper pool_mapper(process_handle, memory_pools, memory_pool_count,
+                           behaviour.IsMemoryForceMappingEnabled());
+
+    std::span<const SinkInfoBase::InParameter> in_params{
+        reinterpret_cast<const SinkInfoBase::InParameter*>(input), memory_pool_count};
+    std::span<SinkInfoBase::OutStatus> out_params{
+        reinterpret_cast<SinkInfoBase::OutStatus*>(output), memory_pool_count};
+
+    const auto sink_count{sink_context.GetCount()};
+
+    for (u32 i = 0; i < sink_count; i++) {
+        const auto& params{in_params[i]};
+        auto sink_info{sink_context.GetInfo(i)};
+
+        if (sink_info->GetType() != params.type) {
+            sink_info->CleanUp();
+            switch (params.type) {
+            case SinkInfoBase::Type::Invalid:
+                std::construct_at<SinkInfoBase>(reinterpret_cast<SinkInfoBase*>(sink_info));
+                break;
+            case SinkInfoBase::Type::DeviceSink:
+                std::construct_at<DeviceSinkInfo>(reinterpret_cast<DeviceSinkInfo*>(sink_info));
+                break;
+            case SinkInfoBase::Type::CircularBufferSink:
+                std::construct_at<CircularBufferSinkInfo>(
+                    reinterpret_cast<CircularBufferSinkInfo*>(sink_info));
+                break;
+            default:
+                LOG_ERROR(Service_Audio, "Invalid sink type {}", static_cast<u32>(params.type));
+                break;
+            }
+        }
+
+        BehaviorInfo::ErrorInfo error_info{};
+        sink_info->Update(error_info, out_params[i], params, pool_mapper);
+
+        if (error_info.error_code.IsError()) {
+            behaviour.AppendError(error_info);
+        }
+    }
+
+    const auto consumed_input_size{sink_count *
+                                   static_cast<u32>(sizeof(SinkInfoBase::InParameter))};
+    const auto consumed_output_size{sink_count * static_cast<u32>(sizeof(SinkInfoBase::OutStatus))};
+    if (consumed_input_size != in_header->sinks_size) {
+        LOG_ERROR(Service_Audio, "Consumed an incorrect sinks size, header size={}, consumed={}",
+                  in_header->sinks_size, consumed_input_size);
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    input += consumed_input_size;
+    output += consumed_output_size;
+    out_header->sinks_size = consumed_output_size;
+    out_header->size += consumed_output_size;
+
+    return ResultSuccess;
+}
+
+Result InfoUpdater::UpdateMemoryPools(std::span<MemoryPoolInfo> memory_pools,
+                                      const u32 memory_pool_count) {
+    PoolMapper pool_mapper(process_handle, memory_pools, memory_pool_count,
+                           behaviour.IsMemoryForceMappingEnabled());
+    std::span<const MemoryPoolInfo::InParameter> in_params{
+        reinterpret_cast<const MemoryPoolInfo::InParameter*>(input), memory_pool_count};
+    std::span<MemoryPoolInfo::OutStatus> out_params{
+        reinterpret_cast<MemoryPoolInfo::OutStatus*>(output), memory_pool_count};
+
+    for (size_t i = 0; i < memory_pool_count; i++) {
+        auto state{pool_mapper.Update(memory_pools[i], in_params[i], out_params[i])};
+        if (state != MemoryPoolInfo::ResultState::Success &&
+            state != MemoryPoolInfo::ResultState::BadParam &&
+            state != MemoryPoolInfo::ResultState::MapFailed &&
+            state != MemoryPoolInfo::ResultState::InUse) {
+            LOG_WARNING(Service_Audio, "Invalid ResultState from updating memory pools");
+            return Service::Audio::ERR_INVALID_UPDATE_DATA;
+        }
+    }
+
+    const auto consumed_input_size{memory_pool_count *
+                                   static_cast<u32>(sizeof(MemoryPoolInfo::InParameter))};
+    const auto consumed_output_size{memory_pool_count *
+                                    static_cast<u32>(sizeof(MemoryPoolInfo::OutStatus))};
+    if (consumed_input_size != in_header->memory_pool_size) {
+        LOG_ERROR(Service_Audio,
+                  "Consumed an incorrect memory pool size, header size={}, consumed={}",
+                  in_header->memory_pool_size, consumed_input_size);
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    input += consumed_input_size;
+    output += consumed_output_size;
+    out_header->memory_pool_size = consumed_output_size;
+    out_header->size += consumed_output_size;
+    return ResultSuccess;
+}
+
+Result InfoUpdater::UpdatePerformanceBuffer(std::span<u8> performance_output,
+                                            const u64 performance_output_size,
+                                            PerformanceManager* performance_manager) {
+    auto in_params{reinterpret_cast<const PerformanceManager::InParameter*>(input)};
+    auto out_params{reinterpret_cast<PerformanceManager::OutStatus*>(output)};
+
+    if (performance_manager != nullptr) {
+        out_params->history_size =
+            performance_manager->CopyHistories(performance_output.data(), performance_output_size);
+        performance_manager->SetDetailTarget(in_params->target_node_id);
+    } else {
+        out_params->history_size = 0;
+    }
+
+    const auto consumed_input_size{static_cast<u32>(sizeof(PerformanceManager::InParameter))};
+    const auto consumed_output_size{static_cast<u32>(sizeof(PerformanceManager::OutStatus))};
+    if (consumed_input_size != in_header->performance_buffer_size) {
+        LOG_ERROR(Service_Audio,
+                  "Consumed an incorrect performance size, header size={}, consumed={}",
+                  in_header->performance_buffer_size, consumed_input_size);
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    input += consumed_input_size;
+    output += consumed_output_size;
+    out_header->performance_buffer_size = consumed_output_size;
+    out_header->size += consumed_output_size;
+    return ResultSuccess;
+}
+
+Result InfoUpdater::UpdateBehaviorInfo(BehaviorInfo& behaviour_) {
+    const auto in_params{reinterpret_cast<const BehaviorInfo::InParameter*>(input)};
+
+    if (!CheckValidRevision(in_params->revision)) {
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    if (in_params->revision != behaviour_.GetUserRevision()) {
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    behaviour_.ClearError();
+    behaviour_.UpdateFlags(in_params->flags);
+
+    if (in_header->behaviour_size != sizeof(BehaviorInfo::InParameter)) {
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    input += sizeof(BehaviorInfo::InParameter);
+    return ResultSuccess;
+}
+
+Result InfoUpdater::UpdateErrorInfo(BehaviorInfo& behaviour_) {
+    auto out_params{reinterpret_cast<BehaviorInfo::OutStatus*>(output)};
+    behaviour_.CopyErrorInfo(out_params->errors, out_params->error_count);
+
+    const auto consumed_output_size{static_cast<u32>(sizeof(BehaviorInfo::OutStatus))};
+
+    output += consumed_output_size;
+    out_header->behaviour_size = consumed_output_size;
+    out_header->size += consumed_output_size;
+    return ResultSuccess;
+}
+
+Result InfoUpdater::UpdateSplitterInfo(SplitterContext& splitter_context) {
+    u32 consumed_size{0};
+    if (!splitter_context.Update(input, consumed_size)) {
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+
+    input += consumed_size;
+
+    return ResultSuccess;
+}
+
+Result InfoUpdater::UpdateRendererInfo(const u64 elapsed_frames) {
+    struct RenderInfo {
+        /* 0x00 */ u64 frames_elapsed;
+        /* 0x08 */ char unk08[0x8];
+    };
+    static_assert(sizeof(RenderInfo) == 0x10, "RenderInfo has the wrong size!");
+
+    auto out_params{reinterpret_cast<RenderInfo*>(output)};
+    out_params->frames_elapsed = elapsed_frames;
+
+    const auto consumed_output_size{static_cast<u32>(sizeof(RenderInfo))};
+
+    output += consumed_output_size;
+    out_header->render_info_size = consumed_output_size;
+    out_header->size += consumed_output_size;
+
+    return ResultSuccess;
+}
+
+Result InfoUpdater::CheckConsumedSize() {
+    if (CpuAddr(input) - CpuAddr(input_origin.data()) != expected_input_size) {
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    } else if (CpuAddr(output) - CpuAddr(output_origin.data()) != expected_output_size) {
+        return Service::Audio::ERR_INVALID_UPDATE_DATA;
+    }
+    return ResultSuccess;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/behavior/info_updater.h

@@ -0,0 +1,205 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <span>
+
+#include "common/common_types.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace AudioCore::AudioRenderer {
+class BehaviorInfo;
+class VoiceContext;
+class MixContext;
+class SinkContext;
+class SplitterContext;
+class EffectContext;
+class MemoryPoolInfo;
+class PerformanceManager;
+
+class InfoUpdater {
+    struct UpdateDataHeader {
+        explicit UpdateDataHeader(u32 revision_) : revision{revision_} {}
+
+        /* 0x00 */ u32 revision;
+        /* 0x04 */ u32 behaviour_size{};
+        /* 0x08 */ u32 memory_pool_size{};
+        /* 0x0C */ u32 voices_size{};
+        /* 0x10 */ u32 voice_resources_size{};
+        /* 0x14 */ u32 effects_size{};
+        /* 0x18 */ u32 mix_size{};
+        /* 0x1C */ u32 sinks_size{};
+        /* 0x20 */ u32 performance_buffer_size{};
+        /* 0x24 */ char unk24[4];
+        /* 0x28 */ u32 render_info_size{};
+        /* 0x2C */ char unk2C[0x10];
+        /* 0x3C */ u32 size{sizeof(UpdateDataHeader)};
+    };
+    static_assert(sizeof(UpdateDataHeader) == 0x40, "UpdateDataHeader has the wrong size!");
+
+public:
+    explicit InfoUpdater(std::span<const u8> input, std::span<u8> output, u32 process_handle,
+                         BehaviorInfo& behaviour);
+
+    /**
+     * Update the voice channel resources.
+     *
+     * @param voice_context - Voice context to update.
+     * @return Result code.
+     */
+    Result UpdateVoiceChannelResources(VoiceContext& voice_context);
+
+    /**
+     * Update voices.
+     *
+     * @param voice_context     - Voice context to update.
+     * @param memory_pools      - Memory pools to use for these voices.
+     * @param memory_pool_count - Number of memory pools.
+     * @return Result code.
+     */
+    Result UpdateVoices(VoiceContext& voice_context, std::span<MemoryPoolInfo> memory_pools,
+                        u32 memory_pool_count);
+
+    /**
+     * Update effects.
+     *
+     * @param effect_context    - Effect context to update.
+     * @param renderer_active   - Whether the AudioRenderer is active.
+     * @param memory_pools      - Memory pools to use for these voices.
+     * @param memory_pool_count - Number of memory pools.
+     * @return Result code.
+     */
+    Result UpdateEffects(EffectContext& effect_context, bool renderer_active,
+                         std::span<MemoryPoolInfo> memory_pools, u32 memory_pool_count);
+
+    /**
+     * Update mixes.
+     *
+     * @param mix_context       - Mix context to update.
+     * @param mix_buffer_count  - Number of mix buffers.
+     * @param effect_context    - Effect context to update effort order.
+     * @param splitter_context  - Splitter context for the mixes.
+     * @return Result code.
+     */
+    Result UpdateMixes(MixContext& mix_context, u32 mix_buffer_count, EffectContext& effect_context,
+                       SplitterContext& splitter_context);
+
+    /**
+     * Update sinks.
+     *
+     * @param sink_context      - Sink context to update.
+     * @param memory_pools      - Memory pools to use for these voices.
+     * @param memory_pool_count - Number of memory pools.
+     * @return Result code.
+     */
+    Result UpdateSinks(SinkContext& sink_context, std::span<MemoryPoolInfo> memory_pools,
+                       u32 memory_pool_count);
+
+    /**
+     * Update memory pools.
+     *
+     * @param memory_pools      - Memory pools to use for these voices.
+     * @param memory_pool_count - Number of memory pools.
+     * @return Result code.
+     */
+    Result UpdateMemoryPools(std::span<MemoryPoolInfo> memory_pools, u32 memory_pool_count);
+
+    /**
+     * Update the performance buffer.
+     *
+     * @param output              - Output buffer for performance metrics.
+     * @param output_size         - Output buffer size.
+     * @param performance_manager - Performance manager..
+     * @return Result code.
+     */
+    Result UpdatePerformanceBuffer(std::span<u8> output, u64 output_size,
+                                   PerformanceManager* performance_manager);
+
+    /**
+     * Update behaviour.
+     *
+     * @param behaviour - Behaviour to update.
+     * @return Result code.
+     */
+    Result UpdateBehaviorInfo(BehaviorInfo& behaviour);
+
+    /**
+     * Update errors.
+     *
+     * @param behaviour - Behaviour to update.
+     * @return Result code.
+     */
+    Result UpdateErrorInfo(BehaviorInfo& behaviour);
+
+    /**
+     * Update splitter.
+     *
+     * @param splitter_context - Splitter context to update.
+     * @return Result code.
+     */
+    Result UpdateSplitterInfo(SplitterContext& splitter_context);
+
+    /**
+     * Update renderer info.
+     *
+     * @param elapsed_frames - Number of elapsed frames.
+     * @return Result code.
+     */
+    Result UpdateRendererInfo(u64 elapsed_frames);
+
+    /**
+     * Check that the input.output sizes match their expected values.
+     *
+     * @return Result code.
+     */
+    Result CheckConsumedSize();
+
+private:
+    /**
+     * Update effects version 1.
+     *
+     * @param effect_context    - Effect context to update.
+     * @param renderer_active   - Is the AudioRenderer active?
+     * @param memory_pools      - Memory pools to use for these voices.
+     * @param memory_pool_count - Number of memory pools.
+     * @return Result code.
+     */
+    Result UpdateEffectsVersion1(EffectContext& effect_context, bool renderer_active,
+                                 std::span<MemoryPoolInfo> memory_pools, u32 memory_pool_count);
+
+    /**
+     * Update effects version 2.
+     *
+     * @param effect_context    - Effect context to update.
+     * @param renderer_active   - Is the AudioRenderer active?
+     * @param memory_pools      - Memory pools to use for these voices.
+     * @param memory_pool_count - Number of memory pools.
+     * @return Result code.
+     */
+    Result UpdateEffectsVersion2(EffectContext& effect_context, bool renderer_active,
+                                 std::span<MemoryPoolInfo> memory_pools, u32 memory_pool_count);
+
+    /// Input buffer
+    u8 const* input;
+    /// Input buffer start
+    std::span<const u8> input_origin;
+    /// Output buffer start
+    u8* output;
+    /// Output buffer start
+    std::span<u8> output_origin;
+    /// Input header
+    const UpdateDataHeader* in_header;
+    /// Output header
+    UpdateDataHeader* out_header;
+    /// Expected input size, see CheckConsumedSize
+    u64 expected_input_size;
+    /// Expected output size, see CheckConsumedSize
+    u64 expected_output_size;
+    /// Unused
+    u32 process_handle;
+    /// Behaviour
+    BehaviorInfo& behaviour;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/command_buffer.cpp

@@ -0,0 +1,678 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/behavior/behavior_info.h"
+#include "audio_core/renderer/command/command_buffer.h"
+#include "audio_core/renderer/command/command_list_header.h"
+#include "audio_core/renderer/command/command_processing_time_estimator.h"
+#include "audio_core/renderer/effect/effect_biquad_filter_info.h"
+#include "audio_core/renderer/effect/effect_delay_info.h"
+#include "audio_core/renderer/effect/effect_reverb_info.h"
+#include "audio_core/renderer/memory/memory_pool_info.h"
+#include "audio_core/renderer/mix/mix_info.h"
+#include "audio_core/renderer/sink/circular_buffer_sink_info.h"
+#include "audio_core/renderer/sink/device_sink_info.h"
+#include "audio_core/renderer/sink/sink_info_base.h"
+#include "audio_core/renderer/voice/voice_info.h"
+#include "audio_core/renderer/voice/voice_state.h"
+
+namespace AudioCore::AudioRenderer {
+
+template <typename T, CommandId Id>
+T& CommandBuffer::GenerateStart(const s32 node_id) {
+    if (size + sizeof(T) >= command_list.size_bytes()) {
+        LOG_ERROR(
+            Service_Audio,
+            "Attempting to write commands beyond the end of allocated command buffer memory!");
+        UNREACHABLE();
+    }
+
+    auto& cmd{*std::construct_at<T>(reinterpret_cast<T*>(&command_list[size]))};
+
+    cmd.magic = CommandMagic;
+    cmd.enabled = true;
+    cmd.type = Id;
+    cmd.size = sizeof(T);
+    cmd.node_id = node_id;
+
+    return cmd;
+}
+
+template <typename T>
+void CommandBuffer::GenerateEnd(T& cmd) {
+    cmd.estimated_process_time = time_estimator->Estimate(cmd);
+    estimated_process_time += cmd.estimated_process_time;
+    size += sizeof(T);
+    count++;
+}
+
+void CommandBuffer::GeneratePcmInt16Version1Command(const s32 node_id,
+                                                    const MemoryPoolInfo& memory_pool_,
+                                                    VoiceInfo& voice_info,
+                                                    const VoiceState& voice_state,
+                                                    const s16 buffer_count, const s8 channel) {
+    auto& cmd{
+        GenerateStart<PcmInt16DataSourceVersion1Command, CommandId::DataSourcePcmInt16Version1>(
+            node_id)};
+
+    cmd.src_quality = voice_info.src_quality;
+    cmd.output_index = buffer_count + channel;
+    cmd.flags = voice_info.flags & 3;
+    cmd.sample_rate = voice_info.sample_rate;
+    cmd.pitch = voice_info.pitch;
+    cmd.channel_index = channel;
+    cmd.channel_count = voice_info.channel_count;
+
+    for (u32 i = 0; i < MaxWaveBuffers; i++) {
+        voice_info.wavebuffers[i].Copy(cmd.wave_buffers[i]);
+    }
+
+    cmd.voice_state = memory_pool_.Translate(CpuAddr(&voice_state), sizeof(VoiceState));
+
+    GenerateEnd<PcmInt16DataSourceVersion1Command>(cmd);
+}
+
+void CommandBuffer::GeneratePcmInt16Version2Command(const s32 node_id, VoiceInfo& voice_info,
+                                                    const VoiceState& voice_state,
+                                                    const s16 buffer_count, const s8 channel) {
+    auto& cmd{
+        GenerateStart<PcmInt16DataSourceVersion2Command, CommandId::DataSourcePcmInt16Version2>(
+            node_id)};
+
+    cmd.src_quality = voice_info.src_quality;
+    cmd.output_index = buffer_count + channel;
+    cmd.flags = voice_info.flags & 3;
+    cmd.sample_rate = voice_info.sample_rate;
+    cmd.pitch = voice_info.pitch;
+    cmd.channel_index = channel;
+    cmd.channel_count = voice_info.channel_count;
+
+    for (u32 i = 0; i < MaxWaveBuffers; i++) {
+        voice_info.wavebuffers[i].Copy(cmd.wave_buffers[i]);
+    }
+
+    cmd.voice_state = memory_pool->Translate(CpuAddr(&voice_state), sizeof(VoiceState));
+
+    GenerateEnd<PcmInt16DataSourceVersion2Command>(cmd);
+}
+
+void CommandBuffer::GeneratePcmFloatVersion1Command(const s32 node_id,
+                                                    const MemoryPoolInfo& memory_pool_,
+                                                    VoiceInfo& voice_info,
+                                                    const VoiceState& voice_state,
+                                                    const s16 buffer_count, const s8 channel) {
+    auto& cmd{
+        GenerateStart<PcmFloatDataSourceVersion1Command, CommandId::DataSourcePcmFloatVersion1>(
+            node_id)};
+
+    cmd.src_quality = voice_info.src_quality;
+    cmd.output_index = buffer_count + channel;
+    cmd.flags = voice_info.flags & 3;
+    cmd.sample_rate = voice_info.sample_rate;
+    cmd.pitch = voice_info.pitch;
+    cmd.channel_index = channel;
+    cmd.channel_count = voice_info.channel_count;
+
+    for (u32 i = 0; i < MaxWaveBuffers; i++) {
+        voice_info.wavebuffers[i].Copy(cmd.wave_buffers[i]);
+    }
+
+    cmd.voice_state = memory_pool_.Translate(CpuAddr(&voice_state), sizeof(VoiceState));
+
+    GenerateEnd<PcmFloatDataSourceVersion1Command>(cmd);
+}
+
+void CommandBuffer::GeneratePcmFloatVersion2Command(const s32 node_id, VoiceInfo& voice_info,
+                                                    const VoiceState& voice_state,
+                                                    const s16 buffer_count, const s8 channel) {
+    auto& cmd{
+        GenerateStart<PcmFloatDataSourceVersion2Command, CommandId::DataSourcePcmFloatVersion2>(
+            node_id)};
+
+    cmd.src_quality = voice_info.src_quality;
+    cmd.output_index = buffer_count + channel;
+    cmd.flags = voice_info.flags & 3;
+    cmd.sample_rate = voice_info.sample_rate;
+    cmd.pitch = voice_info.pitch;
+    cmd.channel_index = channel;
+    cmd.channel_count = voice_info.channel_count;
+
+    for (u32 i = 0; i < MaxWaveBuffers; i++) {
+        voice_info.wavebuffers[i].Copy(cmd.wave_buffers[i]);
+    }
+
+    cmd.voice_state = memory_pool->Translate(CpuAddr(&voice_state), sizeof(VoiceState));
+
+    GenerateEnd<PcmFloatDataSourceVersion2Command>(cmd);
+}
+
+void CommandBuffer::GenerateAdpcmVersion1Command(const s32 node_id,
+                                                 const MemoryPoolInfo& memory_pool_,
+                                                 VoiceInfo& voice_info,
+                                                 const VoiceState& voice_state,
+                                                 const s16 buffer_count, const s8 channel) {
+    auto& cmd{
+        GenerateStart<AdpcmDataSourceVersion1Command, CommandId::DataSourceAdpcmVersion1>(node_id)};
+
+    cmd.src_quality = voice_info.src_quality;
+    cmd.output_index = buffer_count + channel;
+    cmd.flags = voice_info.flags & 3;
+    cmd.sample_rate = voice_info.sample_rate;
+    cmd.pitch = voice_info.pitch;
+
+    for (u32 i = 0; i < MaxWaveBuffers; i++) {
+        voice_info.wavebuffers[i].Copy(cmd.wave_buffers[i]);
+    }
+
+    cmd.voice_state = memory_pool_.Translate(CpuAddr(&voice_state), sizeof(VoiceState));
+    cmd.data_address = voice_info.data_address.GetReference(true);
+    cmd.data_size = voice_info.data_address.GetSize();
+
+    GenerateEnd<AdpcmDataSourceVersion1Command>(cmd);
+}
+
+void CommandBuffer::GenerateAdpcmVersion2Command(const s32 node_id, VoiceInfo& voice_info,
+                                                 const VoiceState& voice_state,
+                                                 const s16 buffer_count, const s8 channel) {
+    auto& cmd{
+        GenerateStart<AdpcmDataSourceVersion2Command, CommandId::DataSourceAdpcmVersion2>(node_id)};
+
+    cmd.src_quality = voice_info.src_quality;
+    cmd.output_index = buffer_count + channel;
+    cmd.flags = voice_info.flags & 3;
+    cmd.sample_rate = voice_info.sample_rate;
+    cmd.pitch = voice_info.pitch;
+    cmd.channel_index = channel;
+    cmd.channel_count = voice_info.channel_count;
+
+    for (u32 i = 0; i < MaxWaveBuffers; i++) {
+        voice_info.wavebuffers[i].Copy(cmd.wave_buffers[i]);
+    }
+
+    cmd.voice_state = memory_pool->Translate(CpuAddr(&voice_state), sizeof(VoiceState));
+    cmd.data_address = voice_info.data_address.GetReference(true);
+    cmd.data_size = voice_info.data_address.GetSize();
+
+    GenerateEnd<AdpcmDataSourceVersion2Command>(cmd);
+}
+
+void CommandBuffer::GenerateVolumeCommand(const s32 node_id, const s16 buffer_offset,
+                                          const s16 input_index, const f32 volume,
+                                          const u8 precision) {
+    auto& cmd{GenerateStart<VolumeCommand, CommandId::Volume>(node_id)};
+
+    cmd.precision = precision;
+    cmd.input_index = buffer_offset + input_index;
+    cmd.output_index = buffer_offset + input_index;
+    cmd.volume = volume;
+
+    GenerateEnd<VolumeCommand>(cmd);
+}
+
+void CommandBuffer::GenerateVolumeRampCommand(const s32 node_id, VoiceInfo& voice_info,
+                                              const s16 buffer_count, const u8 precision) {
+    auto& cmd{GenerateStart<VolumeRampCommand, CommandId::VolumeRamp>(node_id)};
+
+    cmd.input_index = buffer_count;
+    cmd.output_index = buffer_count;
+    cmd.prev_volume = voice_info.prev_volume;
+    cmd.volume = voice_info.volume;
+    cmd.precision = precision;
+
+    GenerateEnd<VolumeRampCommand>(cmd);
+}
+
+void CommandBuffer::GenerateBiquadFilterCommand(const s32 node_id, VoiceInfo& voice_info,
+                                                const VoiceState& voice_state,
+                                                const s16 buffer_count, const s8 channel,
+                                                const u32 biquad_index,
+                                                const bool use_float_processing) {
+    auto& cmd{GenerateStart<BiquadFilterCommand, CommandId::BiquadFilter>(node_id)};
+
+    cmd.input = buffer_count + channel;
+    cmd.output = buffer_count + channel;
+
+    cmd.biquad = voice_info.biquads[biquad_index];
+
+    cmd.state = memory_pool->Translate(CpuAddr(voice_state.biquad_states[biquad_index].data()),
+                                       MaxBiquadFilters * sizeof(VoiceState::BiquadFilterState));
+
+    cmd.needs_init = !voice_info.biquad_initialized[biquad_index];
+    cmd.use_float_processing = use_float_processing;
+
+    GenerateEnd<BiquadFilterCommand>(cmd);
+}
+
+void CommandBuffer::GenerateBiquadFilterCommand(const s32 node_id, EffectInfoBase& effect_info,
+                                                const s16 buffer_offset, const s8 channel,
+                                                const bool needs_init,
+                                                const bool use_float_processing) {
+    auto& cmd{GenerateStart<BiquadFilterCommand, CommandId::BiquadFilter>(node_id)};
+
+    const auto& parameter{
+        *reinterpret_cast<BiquadFilterInfo::ParameterVersion1*>(effect_info.GetParameter())};
+    const auto state{
+        reinterpret_cast<VoiceState::BiquadFilterState*>(effect_info.GetStateBuffer())};
+
+    cmd.input = buffer_offset + parameter.inputs[channel];
+    cmd.output = buffer_offset + parameter.outputs[channel];
+
+    cmd.biquad.b = parameter.b;
+    cmd.biquad.a = parameter.a;
+
+    cmd.state = memory_pool->Translate(CpuAddr(state),
+                                       MaxBiquadFilters * sizeof(VoiceState::BiquadFilterState));
+
+    cmd.needs_init = needs_init;
+    cmd.use_float_processing = use_float_processing;
+
+    GenerateEnd<BiquadFilterCommand>(cmd);
+}
+
+void CommandBuffer::GenerateMixCommand(const s32 node_id, const s16 input_index,
+                                       const s16 output_index, const s16 buffer_offset,
+                                       const f32 volume, const u8 precision) {
+    auto& cmd{GenerateStart<MixCommand, CommandId::Mix>(node_id)};
+
+    cmd.input_index = input_index;
+    cmd.output_index = output_index;
+    cmd.volume = volume;
+    cmd.precision = precision;
+
+    GenerateEnd<MixCommand>(cmd);
+}
+
+void CommandBuffer::GenerateMixRampCommand(const s32 node_id,
+                                           [[maybe_unused]] const s16 buffer_count,
+                                           const s16 input_index, const s16 output_index,
+                                           const f32 volume, const f32 prev_volume,
+                                           const CpuAddr prev_samples, const u8 precision) {
+    if (volume == 0.0f && prev_volume == 0.0f) {
+        return;
+    }
+
+    auto& cmd{GenerateStart<MixRampCommand, CommandId::MixRamp>(node_id)};
+
+    cmd.input_index = input_index;
+    cmd.output_index = output_index;
+    cmd.prev_volume = prev_volume;
+    cmd.volume = volume;
+    cmd.previous_sample = prev_samples;
+    cmd.precision = precision;
+
+    GenerateEnd<MixRampCommand>(cmd);
+}
+
+void CommandBuffer::GenerateMixRampGroupedCommand(const s32 node_id, const s16 buffer_count,
+                                                  const s16 input_index, s16 output_index,
+                                                  std::span<const f32> volumes,
+                                                  std::span<const f32> prev_volumes,
+                                                  const CpuAddr prev_samples, const u8 precision) {
+    auto& cmd{GenerateStart<MixRampGroupedCommand, CommandId::MixRampGrouped>(node_id)};
+
+    cmd.buffer_count = buffer_count;
+
+    for (s32 i = 0; i < buffer_count; i++) {
+        cmd.inputs[i] = input_index;
+        cmd.outputs[i] = output_index++;
+        cmd.prev_volumes[i] = prev_volumes[i];
+        cmd.volumes[i] = volumes[i];
+    }
+
+    cmd.previous_samples = prev_samples;
+    cmd.precision = precision;
+
+    GenerateEnd<MixRampGroupedCommand>(cmd);
+}
+
+void CommandBuffer::GenerateDepopPrepareCommand(const s32 node_id, const VoiceState& voice_state,
+                                                std::span<const s32> buffer, const s16 buffer_count,
+                                                s16 buffer_offset, const bool was_playing) {
+    auto& cmd{GenerateStart<DepopPrepareCommand, CommandId::DepopPrepare>(node_id)};
+
+    cmd.enabled = was_playing;
+
+    for (u32 i = 0; i < MaxMixBuffers; i++) {
+        cmd.inputs[i] = buffer_offset++;
+    }
+
+    cmd.previous_samples = memory_pool->Translate(CpuAddr(voice_state.previous_samples.data()),
+                                                  MaxMixBuffers * sizeof(s32));
+    cmd.buffer_count = buffer_count;
+    cmd.depop_buffer = memory_pool->Translate(CpuAddr(buffer.data()), buffer_count * sizeof(s32));
+
+    GenerateEnd<DepopPrepareCommand>(cmd);
+}
+
+void CommandBuffer::GenerateDepopForMixBuffersCommand(const s32 node_id, const MixInfo& mix_info,
+                                                      std::span<const s32> depop_buffer) {
+    auto& cmd{GenerateStart<DepopForMixBuffersCommand, CommandId::DepopForMixBuffers>(node_id)};
+
+    cmd.input = mix_info.buffer_offset;
+    cmd.count = mix_info.buffer_count;
+    cmd.decay = mix_info.sample_rate == TargetSampleRate ? 0.96218872f : 0.94369507f;
+    cmd.depop_buffer =
+        memory_pool->Translate(CpuAddr(depop_buffer.data()), mix_info.buffer_count * sizeof(s32));
+
+    GenerateEnd<DepopForMixBuffersCommand>(cmd);
+}
+
+void CommandBuffer::GenerateDelayCommand(const s32 node_id, EffectInfoBase& effect_info,
+                                         const s16 buffer_offset) {
+    auto& cmd{GenerateStart<DelayCommand, CommandId::Delay>(node_id)};
+
+    const auto& parameter{
+        *reinterpret_cast<DelayInfo::ParameterVersion1*>(effect_info.GetParameter())};
+    const auto state{effect_info.GetStateBuffer()};
+
+    if (IsChannelCountValid(parameter.channel_count)) {
+        const auto state_buffer{memory_pool->Translate(CpuAddr(state), sizeof(DelayInfo::State))};
+        if (state_buffer) {
+            for (s8 channel = 0; channel < parameter.channel_count; channel++) {
+                cmd.inputs[channel] = static_cast<s8>(buffer_offset + parameter.inputs[channel]);
+                cmd.outputs[channel] = static_cast<s8>(buffer_offset + parameter.outputs[channel]);
+            }
+
+            cmd.parameter = parameter;
+            cmd.effect_enabled = effect_info.IsEnabled();
+            cmd.state = state_buffer;
+            cmd.workbuffer = effect_info.GetWorkbuffer(-1);
+        }
+    }
+
+    GenerateEnd<DelayCommand>(cmd);
+}
+
+void CommandBuffer::GenerateUpsampleCommand(const s32 node_id, const s16 buffer_offset,
+                                            UpsamplerInfo& upsampler_info, const u32 input_count,
+                                            std::span<const s8> inputs, const s16 buffer_count,
+                                            const u32 sample_count_, const u32 sample_rate_) {
+    auto& cmd{GenerateStart<UpsampleCommand, CommandId::Upsample>(node_id)};
+
+    cmd.samples_buffer = memory_pool->Translate(upsampler_info.samples_pos,
+                                                upsampler_info.sample_count * sizeof(s32));
+    cmd.inputs = memory_pool->Translate(CpuAddr(upsampler_info.inputs.data()), MaxChannels);
+    cmd.buffer_count = buffer_count;
+    cmd.unk_20 = 0;
+    cmd.source_sample_count = sample_count_;
+    cmd.source_sample_rate = sample_rate_;
+
+    upsampler_info.input_count = input_count;
+    for (u32 i = 0; i < input_count; i++) {
+        upsampler_info.inputs[i] = buffer_offset + inputs[i];
+    }
+
+    cmd.upsampler_info = memory_pool->Translate(CpuAddr(&upsampler_info), sizeof(UpsamplerInfo));
+
+    GenerateEnd<UpsampleCommand>(cmd);
+}
+
+void CommandBuffer::GenerateDownMix6chTo2chCommand(const s32 node_id, std::span<const s8> inputs,
+                                                   const s16 buffer_offset,
+                                                   std::span<const f32> downmix_coeff) {
+    auto& cmd{GenerateStart<DownMix6chTo2chCommand, CommandId::DownMix6chTo2ch>(node_id)};
+
+    for (u32 i = 0; i < MaxChannels; i++) {
+        cmd.inputs[i] = buffer_offset + inputs[i];
+        cmd.outputs[i] = buffer_offset + inputs[i];
+    }
+
+    for (u32 i = 0; i < 4; i++) {
+        cmd.down_mix_coeff[i] = downmix_coeff[i];
+    }
+
+    GenerateEnd<DownMix6chTo2chCommand>(cmd);
+}
+
+void CommandBuffer::GenerateAuxCommand(const s32 node_id, EffectInfoBase& effect_info,
+                                       const s16 input_index, const s16 output_index,
+                                       const s16 buffer_offset, const u32 update_count,
+                                       const u32 count_max, const u32 write_offset) {
+    auto& cmd{GenerateStart<AuxCommand, CommandId::Aux>(node_id)};
+
+    if (effect_info.GetSendBuffer() != 0 && effect_info.GetReturnBuffer() != 0) {
+        cmd.input = buffer_offset + input_index;
+        cmd.output = buffer_offset + output_index;
+        cmd.send_buffer_info = effect_info.GetSendBufferInfo();
+        cmd.send_buffer = effect_info.GetSendBuffer();
+        cmd.return_buffer_info = effect_info.GetReturnBufferInfo();
+        cmd.return_buffer = effect_info.GetReturnBuffer();
+        cmd.count_max = count_max;
+        cmd.write_offset = write_offset;
+        cmd.update_count = update_count;
+        cmd.effect_enabled = effect_info.IsEnabled();
+    }
+
+    GenerateEnd<AuxCommand>(cmd);
+}
+
+void CommandBuffer::GenerateDeviceSinkCommand(const s32 node_id, const s16 buffer_offset,
+                                              SinkInfoBase& sink_info, const u32 session_id,
+                                              std::span<s32> samples_buffer) {
+    auto& cmd{GenerateStart<DeviceSinkCommand, CommandId::DeviceSink>(node_id)};
+    const auto& parameter{
+        *reinterpret_cast<DeviceSinkInfo::DeviceInParameter*>(sink_info.GetParameter())};
+    auto state{*reinterpret_cast<DeviceSinkInfo::DeviceState*>(sink_info.GetState())};
+
+    cmd.session_id = session_id;
+
+    if (state.upsampler_info != nullptr) {
+        const auto size_{state.upsampler_info->sample_count * parameter.input_count};
+        const auto size_bytes{size_ * sizeof(s32)};
+        const auto addr{memory_pool->Translate(state.upsampler_info->samples_pos, size_bytes)};
+        cmd.sample_buffer = {reinterpret_cast<s32*>(addr),
+                             parameter.input_count * state.upsampler_info->sample_count};
+    } else {
+        cmd.sample_buffer = samples_buffer;
+    }
+
+    cmd.input_count = parameter.input_count;
+    for (u32 i = 0; i < parameter.input_count; i++) {
+        cmd.inputs[i] = buffer_offset + parameter.inputs[i];
+    }
+
+    GenerateEnd<DeviceSinkCommand>(cmd);
+}
+
+void CommandBuffer::GenerateCircularBufferSinkCommand(const s32 node_id, SinkInfoBase& sink_info,
+                                                      const s16 buffer_offset) {
+    auto& cmd{GenerateStart<CircularBufferSinkCommand, CommandId::CircularBufferSink>(node_id)};
+    const auto& parameter{*reinterpret_cast<CircularBufferSinkInfo::CircularBufferInParameter*>(
+        sink_info.GetParameter())};
+    auto state{
+        *reinterpret_cast<CircularBufferSinkInfo::CircularBufferState*>(sink_info.GetState())};
+
+    cmd.input_count = parameter.input_count;
+    for (u32 i = 0; i < parameter.input_count; i++) {
+        cmd.inputs[i] = buffer_offset + parameter.inputs[i];
+    }
+
+    cmd.address = state.address_info.GetReference(true);
+    cmd.size = parameter.size;
+    cmd.pos = state.current_pos;
+
+    GenerateEnd<CircularBufferSinkCommand>(cmd);
+}
+
+void CommandBuffer::GenerateReverbCommand(const s32 node_id, EffectInfoBase& effect_info,
+                                          const s16 buffer_offset,
+                                          const bool long_size_pre_delay_supported) {
+    auto& cmd{GenerateStart<ReverbCommand, CommandId::Reverb>(node_id)};
+
+    const auto& parameter{
+        *reinterpret_cast<ReverbInfo::ParameterVersion2*>(effect_info.GetParameter())};
+    const auto state{effect_info.GetStateBuffer()};
+
+    if (IsChannelCountValid(parameter.channel_count)) {
+        const auto state_buffer{memory_pool->Translate(CpuAddr(state), sizeof(ReverbInfo::State))};
+        if (state_buffer) {
+            for (s8 channel = 0; channel < parameter.channel_count; channel++) {
+                cmd.inputs[channel] = buffer_offset + parameter.inputs[channel];
+                cmd.outputs[channel] = buffer_offset + parameter.outputs[channel];
+            }
+
+            cmd.parameter = parameter;
+            cmd.effect_enabled = effect_info.IsEnabled();
+            cmd.state = state_buffer;
+            cmd.workbuffer = effect_info.GetWorkbuffer(-1);
+            cmd.long_size_pre_delay_supported = long_size_pre_delay_supported;
+        }
+    }
+
+    GenerateEnd<ReverbCommand>(cmd);
+}
+
+void CommandBuffer::GenerateI3dl2ReverbCommand(const s32 node_id, EffectInfoBase& effect_info,
+                                               const s16 buffer_offset) {
+    auto& cmd{GenerateStart<I3dl2ReverbCommand, CommandId::I3dl2Reverb>(node_id)};
+
+    const auto& parameter{
+        *reinterpret_cast<I3dl2ReverbInfo::ParameterVersion1*>(effect_info.GetParameter())};
+    const auto state{effect_info.GetStateBuffer()};
+
+    if (IsChannelCountValid(parameter.channel_count)) {
+        const auto state_buffer{
+            memory_pool->Translate(CpuAddr(state), sizeof(I3dl2ReverbInfo::State))};
+        if (state_buffer) {
+            for (s8 channel = 0; channel < parameter.channel_count; channel++) {
+                cmd.inputs[channel] = buffer_offset + parameter.inputs[channel];
+                cmd.outputs[channel] = buffer_offset + parameter.outputs[channel];
+            }
+
+            cmd.parameter = parameter;
+            cmd.effect_enabled = effect_info.IsEnabled();
+            cmd.state = state_buffer;
+            cmd.workbuffer = effect_info.GetWorkbuffer(-1);
+        }
+    }
+
+    GenerateEnd<I3dl2ReverbCommand>(cmd);
+}
+
+void CommandBuffer::GeneratePerformanceCommand(const s32 node_id, const PerformanceState state,
+                                               const PerformanceEntryAddresses& entry_addresses) {
+    auto& cmd{GenerateStart<PerformanceCommand, CommandId::Performance>(node_id)};
+
+    cmd.state = state;
+    cmd.entry_address = entry_addresses;
+
+    GenerateEnd<PerformanceCommand>(cmd);
+}
+
+void CommandBuffer::GenerateClearMixCommand(const s32 node_id) {
+    auto& cmd{GenerateStart<ClearMixBufferCommand, CommandId::ClearMixBuffer>(node_id)};
+    GenerateEnd<ClearMixBufferCommand>(cmd);
+}
+
+void CommandBuffer::GenerateCopyMixBufferCommand(const s32 node_id, EffectInfoBase& effect_info,
+                                                 const s16 buffer_offset, const s8 channel) {
+    auto& cmd{GenerateStart<CopyMixBufferCommand, CommandId::CopyMixBuffer>(node_id)};
+
+    const auto& parameter{
+        *reinterpret_cast<BiquadFilterInfo::ParameterVersion1*>(effect_info.GetParameter())};
+    cmd.input_index = buffer_offset + parameter.inputs[channel];
+    cmd.output_index = buffer_offset + parameter.outputs[channel];
+
+    GenerateEnd<CopyMixBufferCommand>(cmd);
+}
+
+void CommandBuffer::GenerateLightLimiterCommand(
+    const s32 node_id, const s16 buffer_offset,
+    const LightLimiterInfo::ParameterVersion1& parameter, const LightLimiterInfo::State& state,
+    const bool enabled, const CpuAddr workbuffer) {
+    auto& cmd{GenerateStart<LightLimiterVersion1Command, CommandId::LightLimiterVersion1>(node_id)};
+
+    if (IsChannelCountValid(parameter.channel_count)) {
+        const auto state_buffer{
+            memory_pool->Translate(CpuAddr(&state), sizeof(LightLimiterInfo::State))};
+        if (state_buffer) {
+            for (s8 channel = 0; channel < parameter.channel_count; channel++) {
+                cmd.inputs[channel] = buffer_offset + parameter.inputs[channel];
+                cmd.outputs[channel] = buffer_offset + parameter.outputs[channel];
+            }
+
+            std::memcpy(&cmd.parameter, &parameter, sizeof(LightLimiterInfo::ParameterVersion1));
+            cmd.effect_enabled = enabled;
+            cmd.state = state_buffer;
+            cmd.workbuffer = workbuffer;
+        }
+    }
+
+    GenerateEnd<LightLimiterVersion1Command>(cmd);
+}
+
+void CommandBuffer::GenerateLightLimiterCommand(
+    const s32 node_id, const s16 buffer_offset,
+    const LightLimiterInfo::ParameterVersion2& parameter,
+    const LightLimiterInfo::StatisticsInternal& statistics, const LightLimiterInfo::State& state,
+    const bool enabled, const CpuAddr workbuffer) {
+    auto& cmd{GenerateStart<LightLimiterVersion2Command, CommandId::LightLimiterVersion2>(node_id)};
+    if (IsChannelCountValid(parameter.channel_count)) {
+        const auto state_buffer{
+            memory_pool->Translate(CpuAddr(&state), sizeof(LightLimiterInfo::State))};
+        if (state_buffer) {
+            for (s8 channel = 0; channel < parameter.channel_count; channel++) {
+                cmd.inputs[channel] = buffer_offset + parameter.inputs[channel];
+                cmd.outputs[channel] = buffer_offset + parameter.outputs[channel];
+            }
+
+            cmd.parameter = parameter;
+            cmd.effect_enabled = enabled;
+            cmd.state = state_buffer;
+            if (cmd.parameter.statistics_enabled) {
+                cmd.result_state = memory_pool->Translate(
+                    CpuAddr(&statistics), sizeof(LightLimiterInfo::StatisticsInternal));
+            } else {
+                cmd.result_state = 0;
+            }
+            cmd.workbuffer = workbuffer;
+        }
+    }
+
+    GenerateEnd<LightLimiterVersion2Command>(cmd);
+}
+
+void CommandBuffer::GenerateMultitapBiquadFilterCommand(const s32 node_id, VoiceInfo& voice_info,
+                                                        const VoiceState& voice_state,
+                                                        const s16 buffer_count, const s8 channel) {
+    auto& cmd{GenerateStart<MultiTapBiquadFilterCommand, CommandId::MultiTapBiquadFilter>(node_id)};
+
+    cmd.input = buffer_count + channel;
+    cmd.output = buffer_count + channel;
+    cmd.biquads = voice_info.biquads;
+
+    cmd.states[0] =
+        memory_pool->Translate(CpuAddr(voice_state.biquad_states[0].data()),
+                               MaxBiquadFilters * sizeof(VoiceState::BiquadFilterState));
+    cmd.states[1] =
+        memory_pool->Translate(CpuAddr(voice_state.biquad_states[1].data()),
+                               MaxBiquadFilters * sizeof(VoiceState::BiquadFilterState));
+
+    cmd.needs_init[0] = !voice_info.biquad_initialized[0];
+    cmd.needs_init[1] = !voice_info.biquad_initialized[1];
+    cmd.filter_tap_count = MaxBiquadFilters;
+
+    GenerateEnd<MultiTapBiquadFilterCommand>(cmd);
+}
+
+void CommandBuffer::GenerateCaptureCommand(const s32 node_id, EffectInfoBase& effect_info,
+                                           const s16 input_index, const s16 output_index,
+                                           const s16 buffer_offset, const u32 update_count,
+                                           const u32 count_max, const u32 write_offset) {
+    auto& cmd{GenerateStart<CaptureCommand, CommandId::Capture>(node_id)};
+
+    if (effect_info.GetSendBuffer()) {
+        cmd.input = buffer_offset + input_index;
+        cmd.output = buffer_offset + output_index;
+        cmd.send_buffer_info = effect_info.GetSendBufferInfo();
+        cmd.send_buffer = effect_info.GetSendBuffer();
+        cmd.count_max = count_max;
+        cmd.write_offset = write_offset;
+        cmd.update_count = update_count;
+        cmd.effect_enabled = effect_info.IsEnabled();
+    }
+
+    GenerateEnd<CaptureCommand>(cmd);
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/command_buffer.h

@@ -0,0 +1,457 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <span>
+
+#include "audio_core/renderer/command/commands.h"
+#include "audio_core/renderer/effect/effect_light_limiter_info.h"
+#include "audio_core/renderer/performance/performance_manager.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+struct UpsamplerInfo;
+struct VoiceState;
+class EffectInfoBase;
+class ICommandProcessingTimeEstimator;
+class MixInfo;
+class MemoryPoolInfo;
+class SinkInfoBase;
+class VoiceInfo;
+
+/**
+ * Utility functions to generate and add commands into the current command list.
+ */
+class CommandBuffer {
+public:
+    /**
+     * Generate a PCM s16 version 1 command, adding it to the command list.
+     *
+     * @param node_id      - Node id of the voice this command is generated for.
+     * @param memory_pool  - Memory pool for translating buffer addresses to the DSP.
+     * @param voice_info   - The voice info this command is generated from.
+     * @param voice_state  - The voice state the DSP will use for this command.
+     * @param buffer_count - Number of mix buffers in use,
+     *                       data will be read into this index + channel.
+     * @param channel      - Channel index for this command.
+     */
+    void GeneratePcmInt16Version1Command(s32 node_id, const MemoryPoolInfo& memory_pool,
+                                         VoiceInfo& voice_info, const VoiceState& voice_state,
+                                         s16 buffer_count, s8 channel);
+
+    /**
+     * Generate a PCM s16 version 2 command, adding it to the command list.
+     *
+     * @param node_id      - Node id of the voice this command is generated for.
+     * @param voice_info   - The voice info this command is generated from.
+     * @param voice_state  - The voice state the DSP will use for this command.
+     * @param buffer_count - Number of mix buffers in use,
+     *                       data will be read into this index + channel.
+     * @param channel      - Channel index for this command.
+     */
+    void GeneratePcmInt16Version2Command(s32 node_id, VoiceInfo& voice_info,
+                                         const VoiceState& voice_state, s16 buffer_count,
+                                         s8 channel);
+
+    /**
+     * Generate a PCM f32 version 1 command, adding it to the command list.
+     *
+     * @param node_id      - Node id of the voice this command is generated for.
+     * @param memory_pool  - Memory pool for translating buffer addresses to the DSP.
+     * @param voice_info   - The voice info this command is generated from.
+     * @param voice_state  - The voice state the DSP will use for this command.
+     * @param buffer_count - Number of mix buffers in use,
+     *                       data will be read into this index + channel.
+     * @param channel      - Channel index for this command.
+     */
+    void GeneratePcmFloatVersion1Command(s32 node_id, const MemoryPoolInfo& memory_pool,
+                                         VoiceInfo& voice_info, const VoiceState& voice_state,
+                                         s16 buffer_count, s8 channel);
+
+    /**
+     * Generate a PCM f32 version 2 command, adding it to the command list.
+     *
+     * @param node_id      - Node id of the voice this command is generated for.
+     * @param voice_info   - The voice info this command is generated from.
+     * @param voice_state  - The voice state the DSP will use for this command.
+     * @param buffer_count - Number of mix buffers in use,
+     *                       data will be read into this index + channel.
+     * @param channel      - Channel index for this command.
+     */
+    void GeneratePcmFloatVersion2Command(s32 node_id, VoiceInfo& voice_info,
+                                         const VoiceState& voice_state, s16 buffer_count,
+                                         s8 channel);
+
+    /**
+     * Generate an ADPCM version 1 command, adding it to the command list.
+     *
+     * @param node_id      - Node id of the voice this command is generated for.
+     * @param memory_pool  - Memory pool for translating buffer addresses to the DSP.
+     * @param voice_info   - The voice info this command is generated from.
+     * @param voice_state  - The voice state the DSP will use for this command.
+     * @param buffer_count - Number of mix buffers in use,
+     *                       data will be read into this index + channel.
+     * @param channel      - Channel index for this command.
+     */
+    void GenerateAdpcmVersion1Command(s32 node_id, const MemoryPoolInfo& memory_pool,
+                                      VoiceInfo& voice_info, const VoiceState& voice_state,
+                                      s16 buffer_count, s8 channel);
+
+    /**
+     * Generate an ADPCM version 2 command, adding it to the command list.
+     *
+     * @param node_id      - Node id of the voice this command is generated for.
+     * @param voice_info   - The voice info this command is generated from.
+     * @param voice_state  - The voice state the DSP will use for this command.
+     * @param buffer_count - Number of mix buffers in use,
+     *                       data will be read into this index + channel.
+     * @param channel      - Channel index for this command.
+     */
+    void GenerateAdpcmVersion2Command(s32 node_id, VoiceInfo& voice_info,
+                                      const VoiceState& voice_state, s16 buffer_count, s8 channel);
+
+    /**
+     * Generate a volume command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param buffer_offset - Base mix buffer index to generate this command at.
+     * @param input_index   - Channel index and mix buffer offset for this command.
+     * @param volume        - Mix volume added to the input samples.
+     * @param precision     - Number of decimal bits for fixed point operations.
+     */
+    void GenerateVolumeCommand(s32 node_id, s16 buffer_offset, s16 input_index, f32 volume,
+                               u8 precision);
+
+    /**
+     * Generate a volume ramp command, adding it to the command list.
+     *
+     * @param node_id      - Node id of the voice this command is generated for.
+     * @param voice_info   - The voice info this command takes its volumes from.
+     * @param buffer_count - Number of active mix buffers, command will generate at this index.
+     * @param precision    - Number of decimal bits for fixed point operations.
+     */
+    void GenerateVolumeRampCommand(s32 node_id, VoiceInfo& voice_info, s16 buffer_count,
+                                   u8 precision);
+
+    /**
+     * Generate a biquad filter command from a voice, adding it to the command list.
+     *
+     * @param node_id              - Node id of the voice this command is generated for.
+     * @param voice_info           - The voice info this command takes biquad parameters from.
+     * @param voice_state          - Used by the AudioRenderer to track previous samples.
+     * @param buffer_count         - Number of active mix buffers,
+     *                               command will generate at this index + channel.
+     * @param channel              - Channel index for this filter to work on.
+     * @param biquad_index         - Which biquad filter to use for this command (0-1).
+     * @param use_float_processing - Should int or float processing be used?
+     */
+    void GenerateBiquadFilterCommand(s32 node_id, VoiceInfo& voice_info,
+                                     const VoiceState& voice_state, s16 buffer_count, s8 channel,
+                                     u32 biquad_index, bool use_float_processing);
+
+    /**
+     * Generate a biquad filter effect command, adding it to the command list.
+     *
+     * @param node_id              - Node id of the voice this command is generated for.
+     * @param effect_info          - The effect info this command takes biquad parameters from.
+     * @param buffer_offset        - Mix buffer offset this command will use,
+     *                               command will generate at this index + channel.
+     * @param channel              - Channel index for this filter to work on.
+     * @param needs_init           - True if the biquad state needs initialisation.
+     * @param use_float_processing - Should int or float processing be used?
+     */
+    void GenerateBiquadFilterCommand(s32 node_id, EffectInfoBase& effect_info, s16 buffer_offset,
+                                     s8 channel, bool needs_init, bool use_float_processing);
+
+    /**
+     * Generate a mix command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param input_index   - Input mix buffer index for this command.
+     *                        Added to the buffer offset.
+     * @param output_index  - Output mix buffer index for this command.
+     *                        Added to the buffer offset.
+     * @param buffer_offset - Mix buffer offset this command will use.
+     * @param volume        - Volume to be applied to the input.
+     * @param precision     - Number of decimal bits for fixed point operations.
+     */
+    void GenerateMixCommand(s32 node_id, s16 input_index, s16 output_index, s16 buffer_offset,
+                            f32 volume, u8 precision);
+
+    /**
+     * Generate a mix ramp command, adding it to the command list.
+     *
+     * @param node_id      - Node id of the voice this command is generated for.
+     * @param buffer_count - Number of active mix buffers.
+     * @param input_index  - Input mix buffer index for this command.
+     *                       Added to buffer_count.
+     * @param output_index - Output mix buffer index for this command.
+     *                       Added to buffer_count.
+     * @param volume       - Current mix volume used for calculating the ramp.
+     * @param prev_volume  - Previous mix volume, used for calculating the ramp,
+     *                       also applied to the input.
+     * @param precision    - Number of decimal bits for fixed point operations.
+     */
+    void GenerateMixRampCommand(s32 node_id, s16 buffer_count, s16 input_index, s16 output_index,
+                                f32 volume, f32 prev_volume, CpuAddr prev_samples, u8 precision);
+
+    /**
+     * Generate a mix ramp grouped command, adding it to the command list.
+     *
+     * @param node_id      - Node id of the voice this command is generated for.
+     * @param buffer_count - Number of active mix buffers.
+     * @param input_index  - Input mix buffer index for this command.
+     *                       Added to buffer_count.
+     * @param output_index - Output mix buffer index for this command.
+     *                       Added to buffer_count.
+     * @param volumes      - Current mix volumes used for calculating the ramp.
+     * @param prev_volumes - Previous mix volumes, used for calculating the ramp,
+     *                       also applied to the input.
+     * @param precision    - Number of decimal bits for fixed point operations.
+     */
+    void GenerateMixRampGroupedCommand(s32 node_id, s16 buffer_count, s16 input_index,
+                                       s16 output_index, std::span<const f32> volumes,
+                                       std::span<const f32> prev_volumes, CpuAddr prev_samples,
+                                       u8 precision);
+
+    /**
+     * Generate a depop prepare command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param voice_state   - State to track the previous depop samples for each mix buffer.
+     * @param buffer        - State to track the current depop samples for each mix buffer.
+     * @param buffer_count  - Number of active mix buffers.
+     * @param buffer_offset - Base mix buffer index to generate the channel depops at.
+     * @param was_playing   - Command only needs to work if the voice was previously playing.
+     */
+    void GenerateDepopPrepareCommand(s32 node_id, const VoiceState& voice_state,
+                                     std::span<const s32> buffer, s16 buffer_count,
+                                     s16 buffer_offset, bool was_playing);
+
+    /**
+     * Generate a depop command, adding it to the command list.
+     *
+     * @param node_id      - Node id of the voice this command is generated for.
+     * @param mix_info     - Mix info to get the buffer count and base offsets from.
+     * @param depop_buffer - Buffer of current depop sample values to be added to the input
+     *                       channels.
+     */
+    void GenerateDepopForMixBuffersCommand(s32 node_id, const MixInfo& mix_info,
+                                           std::span<const s32> depop_buffer);
+
+    /**
+     * Generate a delay command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param effect_info   - Delay effect info to generate this command from.
+     * @param buffer_offset - Base mix buffer offset to apply the apply the delay.
+     */
+    void GenerateDelayCommand(s32 node_id, EffectInfoBase& effect_info, s16 buffer_offset);
+
+    /**
+     * Generate an upsample command, adding it to the command list.
+     *
+     * @param node_id        - Node id of the voice this command is generated for.
+     * @param buffer_offset  - Base mix buffer offset to upsample.
+     * @param upsampler_info - Upsampler info to control the upsampling.
+     * @param input_count    - Number of input channels to upsample.
+     * @param inputs         - Input mix buffer indexes.
+     * @param buffer_count   - Number of active mix buffers.
+     * @param sample_count   - Source sample count of the input.
+     * @param sample_rate    - Source sample rate of the input.
+     */
+    void GenerateUpsampleCommand(s32 node_id, s16 buffer_offset, UpsamplerInfo& upsampler_info,
+                                 u32 input_count, std::span<const s8> inputs, s16 buffer_count,
+                                 u32 sample_count, u32 sample_rate);
+
+    /**
+     * Generate a downmix 6 -> 2 command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param inputs        - Input mix buffer indexes.
+     * @param buffer_offset - Base mix buffer offset of the channels to downmix.
+     * @param downmix_coeff - Downmixing coefficients.
+     */
+    void GenerateDownMix6chTo2chCommand(s32 node_id, std::span<const s8> inputs, s16 buffer_offset,
+                                        std::span<const f32> downmix_coeff);
+
+    /**
+     * Generate an aux buffer command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param effect_info   - Aux effect info to generate this command from.
+     * @param input_index   - Input mix buffer index for this command.
+     *                        Added to buffer_offset.
+     * @param output_index  - Output mix buffer index for this command.
+     *                        Added to buffer_offset.
+     * @param buffer_offset - Base mix buffer offset to use.
+     * @param update_count  - Number of samples to write back to the game as updated, can be 0.
+     * @param count_max     - Maximum number of samples to read or write.
+     * @param write_offset  - Current read or write offset within the buffer.
+     */
+    void GenerateAuxCommand(s32 node_id, EffectInfoBase& effect_info, s16 input_index,
+                            s16 output_index, s16 buffer_offset, u32 update_count, u32 count_max,
+                            u32 write_offset);
+
+    /**
+     * Generate a device sink command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param buffer_offset - Base mix buffer offset to use.
+     * @param sink_info     - The sink_info to generate this command from.
+     * @session_id          - System session id this command is generated from.
+     * @samples_buffer      - The buffer to be sent to the sink if upsampling is not used.
+     */
+    void GenerateDeviceSinkCommand(s32 node_id, s16 buffer_offset, SinkInfoBase& sink_info,
+                                   u32 session_id, std::span<s32> samples_buffer);
+
+    /**
+     * Generate a circular buffer sink command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param sink_info     - The sink_info to generate this command from.
+     * @param buffer_offset - Base mix buffer offset to use.
+     */
+    void GenerateCircularBufferSinkCommand(s32 node_id, SinkInfoBase& sink_info, s16 buffer_offset);
+
+    /**
+     * Generate a reverb command, adding it to the command list.
+     *
+     * @param node_id                       - Node id of the voice this command is generated for.
+     * @param effect_info                   - Reverb effect info to generate this command from.
+     * @param buffer_offset                 - Base mix buffer offset to use.
+     * @param long_size_pre_delay_supported - Should a longer pre-delay time be used before reverb
+     *                                        begins?
+     */
+    void GenerateReverbCommand(s32 node_id, EffectInfoBase& effect_info, s16 buffer_offset,
+                               bool long_size_pre_delay_supported);
+
+    /**
+     * Generate an I3DL2 reverb command, adding it to the command list.
+     *
+     * @param node_id                       - Node id of the voice this command is generated for.
+     * @param effect_info                   - I3DL2Reverb effect info to generate this command from.
+     * @param buffer_offset                 - Base mix buffer offset to use.
+     */
+    void GenerateI3dl2ReverbCommand(s32 node_id, EffectInfoBase& effect_info, s16 buffer_offset);
+
+    /**
+     * Generate a performance command, adding it to the command list.
+     *
+     * @param node_id         - Node id of the voice this command is generated for.
+     * @param state           - State of the performance.
+     * @param entry_addresses - The addresses to be filled in by the AudioRenderer.
+     */
+    void GeneratePerformanceCommand(s32 node_id, PerformanceState state,
+                                    const PerformanceEntryAddresses& entry_addresses);
+
+    /**
+     * Generate a clear mix command, adding it to the command list.
+     *
+     * @param node_id         - Node id of the voice this command is generated for.
+     */
+    void GenerateClearMixCommand(s32 node_id);
+
+    /**
+     * Generate a copy mix command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param effect_info   - BiquadFilter effect info to generate this command from.
+     * @param buffer_offset - Base mix buffer offset to use.
+     * @param channel       - Index to the effect's parameters input indexes for this command.
+     */
+    void GenerateCopyMixBufferCommand(s32 node_id, EffectInfoBase& effect_info, s16 buffer_offset,
+                                      s8 channel);
+
+    /**
+     * Generate a light limiter version 1 command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param buffer_offset - Base mix buffer offset to use.
+     * @param parameter     - Effect parameter to generate from.
+     * @param state         - State used by the AudioRenderer between commands.
+     * @param enabled       - Is this command enabled?
+     * @param workbuffer    - Game-supplied memory for the state.
+     */
+    void GenerateLightLimiterCommand(s32 node_id, s16 buffer_offset,
+                                     const LightLimiterInfo::ParameterVersion1& parameter,
+                                     const LightLimiterInfo::State& state, bool enabled,
+                                     CpuAddr workbuffer);
+
+    /**
+     * Generate a light limiter version 2 command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param buffer_offset - Base mix buffer offset to use.
+     * @param parameter     - Effect parameter to generate from.
+     * @param statistics    - Statistics reported by the AudioRenderer on the limiter's state.
+     * @param state         - State used by the AudioRenderer between commands.
+     * @param enabled       - Is this command enabled?
+     * @param workbuffer    - Game-supplied memory for the state.
+     */
+    void GenerateLightLimiterCommand(s32 node_id, s16 buffer_offset,
+                                     const LightLimiterInfo::ParameterVersion2& parameter,
+                                     const LightLimiterInfo::StatisticsInternal& statistics,
+                                     const LightLimiterInfo::State& state, bool enabled,
+                                     CpuAddr workbuffer);
+
+    /**
+     * Generate a multitap biquad filter command, adding it to the command list.
+     *
+     * @param node_id      - Node id of the voice this command is generated for.
+     * @param voice_info   - The voice info this command takes biquad parameters from.
+     * @param voice_state  - Used by the AudioRenderer to track previous samples.
+     * @param buffer_count - Number of active mix buffers,
+     *                       command will generate at this index + channel.
+     * @param channel      - Channel index for this filter to work on.
+     */
+    void GenerateMultitapBiquadFilterCommand(s32 node_id, VoiceInfo& voice_info,
+                                             const VoiceState& voice_state, s16 buffer_count,
+                                             s8 channel);
+
+    /**
+     * Generate a capture command, adding it to the command list.
+     *
+     * @param node_id       - Node id of the voice this command is generated for.
+     * @param effect_info   - Capture effect info to generate this command from.
+     * @param input_index   - Input mix buffer index for this command.
+     *                        Added to buffer_offset.
+     * @param output_index  - Output mix buffer index for this command (unused).
+     *                        Added to buffer_offset.
+     * @param buffer_offset - Base mix buffer offset to use.
+     * @param update_count  - Number of samples to write back to the game as updated, can be 0.
+     * @param count_max     - Maximum number of samples to read or write.
+     * @param write_offset  - Current read or write offset within the buffer.
+     */
+    void GenerateCaptureCommand(s32 node_id, EffectInfoBase& effect_info, s16 input_index,
+                                s16 output_index, s16 buffer_offset, u32 update_count,
+                                u32 count_max, u32 write_offset);
+
+    /// Command list buffer generated commands will be added to
+    std::span<u8> command_list{};
+    /// Input sample count, unused
+    u32 sample_count{};
+    /// Input sample rate, unused
+    u32 sample_rate{};
+    /// Current size of the command buffer
+    u64 size{};
+    /// Current number of commands added
+    u32 count{};
+    /// Current estimated processing time for all commands
+    u32 estimated_process_time{};
+    /// Used for mapping buffers for the AudioRenderer
+    MemoryPoolInfo* memory_pool{};
+    /// Used for estimating command process times
+    ICommandProcessingTimeEstimator* time_estimator{};
+    /// Used to check which rendering features are currently enabled
+    BehaviorInfo* behavior{};
+
+private:
+    template <typename T, CommandId Id>
+    T& GenerateStart(const s32 node_id);
+    template <typename T>
+    void GenerateEnd(T& cmd);
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/command_generator.cpp

@@ -0,0 +1,783 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/common/audio_renderer_parameter.h"
+#include "audio_core/renderer/behavior/behavior_info.h"
+#include "audio_core/renderer/command/command_buffer.h"
+#include "audio_core/renderer/command/command_generator.h"
+#include "audio_core/renderer/command/command_list_header.h"
+#include "audio_core/renderer/effect/effect_aux_info.h"
+#include "audio_core/renderer/effect/effect_biquad_filter_info.h"
+#include "audio_core/renderer/effect/effect_buffer_mixer_info.h"
+#include "audio_core/renderer/effect/effect_capture_info.h"
+#include "audio_core/renderer/effect/effect_context.h"
+#include "audio_core/renderer/mix/mix_context.h"
+#include "audio_core/renderer/performance/detail_aspect.h"
+#include "audio_core/renderer/performance/entry_aspect.h"
+#include "audio_core/renderer/sink/device_sink_info.h"
+#include "audio_core/renderer/sink/sink_context.h"
+#include "audio_core/renderer/splitter/splitter_context.h"
+#include "audio_core/renderer/voice/voice_context.h"
+#include "common/alignment.h"
+
+namespace AudioCore::AudioRenderer {
+
+CommandGenerator::CommandGenerator(CommandBuffer& command_buffer_,
+                                   const CommandListHeader& command_list_header_,
+                                   const AudioRendererSystemContext& render_context_,
+                                   VoiceContext& voice_context_, MixContext& mix_context_,
+                                   EffectContext& effect_context_, SinkContext& sink_context_,
+                                   SplitterContext& splitter_context_,
+                                   PerformanceManager* performance_manager_)
+    : command_buffer{command_buffer_}, command_header{command_list_header_},
+      render_context{render_context_}, voice_context{voice_context_}, mix_context{mix_context_},
+      effect_context{effect_context_}, sink_context{sink_context_},
+      splitter_context{splitter_context_}, performance_manager{performance_manager_} {
+    command_buffer.GenerateClearMixCommand(InvalidNodeId);
+}
+
+void CommandGenerator::GenerateDataSourceCommand(VoiceInfo& voice_info,
+                                                 const VoiceState& voice_state, const s8 channel) {
+    if (voice_info.mix_id == UnusedMixId) {
+        if (voice_info.splitter_id != UnusedSplitterId) {
+            auto destination{splitter_context.GetDesintationData(voice_info.splitter_id, 0)};
+            u32 dest_id{0};
+            while (destination != nullptr) {
+                if (destination->IsConfigured()) {
+                    auto mix_id{destination->GetMixId()};
+                    if (mix_id < mix_context.GetCount()) {
+                        auto mix_info{mix_context.GetInfo(mix_id)};
+                        command_buffer.GenerateDepopPrepareCommand(
+                            voice_info.node_id, voice_state, render_context.depop_buffer,
+                            mix_info->buffer_count, mix_info->buffer_offset,
+                            voice_info.was_playing);
+                    }
+                }
+                dest_id++;
+                destination = splitter_context.GetDesintationData(voice_info.splitter_id, dest_id);
+            }
+        }
+    } else {
+        auto mix_info{mix_context.GetInfo(voice_info.mix_id)};
+        command_buffer.GenerateDepopPrepareCommand(
+            voice_info.node_id, voice_state, render_context.depop_buffer, mix_info->buffer_count,
+            mix_info->buffer_offset, voice_info.was_playing);
+    }
+
+    if (voice_info.was_playing) {
+        return;
+    }
+
+    if (render_context.behavior->IsWaveBufferVer2Supported()) {
+        switch (voice_info.sample_format) {
+        case SampleFormat::PcmInt16:
+            command_buffer.GeneratePcmInt16Version2Command(
+                voice_info.node_id, voice_info, voice_state, render_context.mix_buffer_count,
+                channel);
+            break;
+        case SampleFormat::PcmFloat:
+            command_buffer.GeneratePcmFloatVersion2Command(
+                voice_info.node_id, voice_info, voice_state, render_context.mix_buffer_count,
+                channel);
+            break;
+        case SampleFormat::Adpcm:
+            command_buffer.GenerateAdpcmVersion2Command(voice_info.node_id, voice_info, voice_state,
+                                                        render_context.mix_buffer_count, channel);
+            break;
+        default:
+            LOG_ERROR(Service_Audio, "Invalid SampleFormat {}",
+                      static_cast<u32>(voice_info.sample_format));
+            break;
+        }
+    } else {
+        switch (voice_info.sample_format) {
+        case SampleFormat::PcmInt16:
+            command_buffer.GeneratePcmInt16Version1Command(
+                voice_info.node_id, *command_buffer.memory_pool, voice_info, voice_state,
+                render_context.mix_buffer_count, channel);
+            break;
+        case SampleFormat::PcmFloat:
+            command_buffer.GeneratePcmFloatVersion1Command(
+                voice_info.node_id, *command_buffer.memory_pool, voice_info, voice_state,
+                render_context.mix_buffer_count, channel);
+            break;
+        case SampleFormat::Adpcm:
+            command_buffer.GenerateAdpcmVersion1Command(
+                voice_info.node_id, *command_buffer.memory_pool, voice_info, voice_state,
+                render_context.mix_buffer_count, channel);
+            break;
+        default:
+            LOG_ERROR(Service_Audio, "Invalid SampleFormat {}",
+                      static_cast<u32>(voice_info.sample_format));
+            break;
+        }
+    }
+}
+
+void CommandGenerator::GenerateVoiceMixCommand(std::span<const f32> mix_volumes,
+                                               std::span<const f32> prev_mix_volumes,
+                                               const VoiceState& voice_state, s16 output_index,
+                                               const s16 buffer_count, const s16 input_index,
+                                               const s32 node_id) {
+    u8 precision{15};
+    if (render_context.behavior->IsVolumeMixParameterPrecisionQ23Supported()) {
+        precision = 23;
+    }
+
+    if (buffer_count > 8) {
+        const auto prev_samples{render_context.memory_pool_info->Translate(
+            CpuAddr(voice_state.previous_samples.data()), buffer_count * sizeof(s32))};
+        command_buffer.GenerateMixRampGroupedCommand(node_id, buffer_count, input_index,
+                                                     output_index, mix_volumes, prev_mix_volumes,
+                                                     prev_samples, precision);
+    } else {
+        for (s16 i = 0; i < buffer_count; i++, output_index++) {
+            const auto prev_samples{render_context.memory_pool_info->Translate(
+                CpuAddr(&voice_state.previous_samples[i]), sizeof(s32))};
+
+            command_buffer.GenerateMixRampCommand(node_id, buffer_count, input_index, output_index,
+                                                  mix_volumes[i], prev_mix_volumes[i], prev_samples,
+                                                  precision);
+        }
+    }
+}
+
+void CommandGenerator::GenerateBiquadFilterCommandForVoice(VoiceInfo& voice_info,
+                                                           const VoiceState& voice_state,
+                                                           const s16 buffer_count, const s8 channel,
+                                                           const s32 node_id) {
+    const bool both_biquads_enabled{voice_info.biquads[0].enabled && voice_info.biquads[1].enabled};
+    const auto use_float_processing{render_context.behavior->UseBiquadFilterFloatProcessing()};
+
+    if (both_biquads_enabled && render_context.behavior->UseMultiTapBiquadFilterProcessing() &&
+        use_float_processing) {
+        command_buffer.GenerateMultitapBiquadFilterCommand(node_id, voice_info, voice_state,
+                                                           buffer_count, channel);
+    } else {
+        for (u32 i = 0; i < MaxBiquadFilters; i++) {
+            if (voice_info.biquads[i].enabled) {
+                command_buffer.GenerateBiquadFilterCommand(node_id, voice_info, voice_state,
+                                                           buffer_count, channel, i,
+                                                           use_float_processing);
+            }
+        }
+    }
+}
+
+void CommandGenerator::GenerateVoiceCommand(VoiceInfo& voice_info) {
+    u8 precision{15};
+    if (render_context.behavior->IsVolumeMixParameterPrecisionQ23Supported()) {
+        precision = 23;
+    }
+
+    for (s8 channel = 0; channel < voice_info.channel_count; channel++) {
+        const auto resource_id{voice_info.channel_resource_ids[channel]};
+        auto& voice_state{voice_context.GetDspSharedState(resource_id)};
+        auto& channel_resource{voice_context.GetChannelResource(resource_id)};
+
+        PerformanceDetailType detail_type{PerformanceDetailType::Invalid};
+        switch (voice_info.sample_format) {
+        case SampleFormat::PcmInt16:
+            detail_type = PerformanceDetailType::Unk1;
+            break;
+        case SampleFormat::PcmFloat:
+            detail_type = PerformanceDetailType::Unk10;
+            break;
+        default:
+            detail_type = PerformanceDetailType::Unk2;
+            break;
+        }
+
+        DetailAspect data_source_detail(*this, PerformanceEntryType::Voice, voice_info.node_id,
+                                        detail_type);
+        GenerateDataSourceCommand(voice_info, voice_state, channel);
+
+        if (data_source_detail.initialized) {
+            command_buffer.GeneratePerformanceCommand(data_source_detail.node_id,
+                                                      PerformanceState::Stop,
+                                                      data_source_detail.performance_entry_address);
+        }
+
+        if (voice_info.was_playing) {
+            voice_info.prev_volume = 0.0f;
+            continue;
+        }
+
+        if (!voice_info.HasAnyConnection()) {
+            continue;
+        }
+
+        DetailAspect biquad_detail_aspect(*this, PerformanceEntryType::Voice, voice_info.node_id,
+                                          PerformanceDetailType::Unk4);
+        GenerateBiquadFilterCommandForVoice(
+            voice_info, voice_state, render_context.mix_buffer_count, channel, voice_info.node_id);
+
+        if (biquad_detail_aspect.initialized) {
+            command_buffer.GeneratePerformanceCommand(
+                biquad_detail_aspect.node_id, PerformanceState::Stop,
+                biquad_detail_aspect.performance_entry_address);
+        }
+
+        DetailAspect volume_ramp_detail_aspect(*this, PerformanceEntryType::Voice,
+                                               voice_info.node_id, PerformanceDetailType::Unk3);
+        command_buffer.GenerateVolumeRampCommand(
+            voice_info.node_id, voice_info, render_context.mix_buffer_count + channel, precision);
+        if (volume_ramp_detail_aspect.initialized) {
+            command_buffer.GeneratePerformanceCommand(
+                volume_ramp_detail_aspect.node_id, PerformanceState::Stop,
+                volume_ramp_detail_aspect.performance_entry_address);
+        }
+
+        voice_info.prev_volume = voice_info.volume;
+
+        if (voice_info.mix_id == UnusedMixId) {
+            if (voice_info.splitter_id != UnusedSplitterId) {
+                auto i{channel};
+                auto destination{splitter_context.GetDesintationData(voice_info.splitter_id, i)};
+                while (destination != nullptr) {
+                    if (destination->IsConfigured()) {
+                        const auto mix_id{destination->GetMixId()};
+                        if (mix_id < mix_context.GetCount() &&
+                            static_cast<s32>(mix_id) != UnusedSplitterId) {
+                            auto mix_info{mix_context.GetInfo(mix_id)};
+                            GenerateVoiceMixCommand(
+                                destination->GetMixVolume(), destination->GetMixVolumePrev(),
+                                voice_state, mix_info->buffer_offset, mix_info->buffer_count,
+                                render_context.mix_buffer_count + channel, voice_info.node_id);
+                            destination->MarkAsNeedToUpdateInternalState();
+                        }
+                    }
+                    i += voice_info.channel_count;
+                    destination = splitter_context.GetDesintationData(voice_info.splitter_id, i);
+                }
+            }
+        } else {
+            DetailAspect volume_mix_detail_aspect(*this, PerformanceEntryType::Voice,
+                                                  voice_info.node_id, PerformanceDetailType::Unk3);
+            auto mix_info{mix_context.GetInfo(voice_info.mix_id)};
+            GenerateVoiceMixCommand(channel_resource.mix_volumes, channel_resource.prev_mix_volumes,
+                                    voice_state, mix_info->buffer_offset, mix_info->buffer_count,
+                                    render_context.mix_buffer_count + channel, voice_info.node_id);
+            if (volume_mix_detail_aspect.initialized) {
+                command_buffer.GeneratePerformanceCommand(
+                    volume_mix_detail_aspect.node_id, PerformanceState::Stop,
+                    volume_mix_detail_aspect.performance_entry_address);
+            }
+
+            channel_resource.prev_mix_volumes = channel_resource.mix_volumes;
+        }
+        voice_info.biquad_initialized[0] = voice_info.biquads[0].enabled;
+        voice_info.biquad_initialized[1] = voice_info.biquads[1].enabled;
+    }
+}
+
+void CommandGenerator::GenerateVoiceCommands() {
+    const auto voice_count{voice_context.GetCount()};
+
+    for (u32 i = 0; i < voice_count; i++) {
+        auto sorted_info{voice_context.GetSortedInfo(i)};
+
+        if (sorted_info->ShouldSkip() || !sorted_info->UpdateForCommandGeneration(voice_context)) {
+            continue;
+        }
+
+        EntryAspect voice_entry_aspect(*this, PerformanceEntryType::Voice, sorted_info->node_id);
+
+        GenerateVoiceCommand(*sorted_info);
+
+        if (voice_entry_aspect.initialized) {
+            command_buffer.GeneratePerformanceCommand(voice_entry_aspect.node_id,
+                                                      PerformanceState::Stop,
+                                                      voice_entry_aspect.performance_entry_address);
+        }
+    }
+
+    splitter_context.UpdateInternalState();
+}
+
+void CommandGenerator::GenerateBufferMixerCommand(const s16 buffer_offset,
+                                                  EffectInfoBase& effect_info, const s32 node_id) {
+    u8 precision{15};
+    if (render_context.behavior->IsVolumeMixParameterPrecisionQ23Supported()) {
+        precision = 23;
+    }
+
+    if (effect_info.IsEnabled()) {
+        const auto& parameter{
+            *reinterpret_cast<BufferMixerInfo::ParameterVersion1*>(effect_info.GetParameter())};
+        for (u32 i = 0; i < parameter.mix_count; i++) {
+            if (parameter.volumes[i] != 0.0f) {
+                command_buffer.GenerateMixCommand(node_id, buffer_offset + parameter.inputs[i],
+                                                  buffer_offset + parameter.outputs[i],
+                                                  buffer_offset, parameter.volumes[i], precision);
+            }
+        }
+    }
+}
+
+void CommandGenerator::GenerateDelayCommand(const s16 buffer_offset, EffectInfoBase& effect_info,
+                                            const s32 node_id) {
+    command_buffer.GenerateDelayCommand(node_id, effect_info, buffer_offset);
+}
+
+void CommandGenerator::GenerateReverbCommand(const s16 buffer_offset, EffectInfoBase& effect_info,
+                                             const s32 node_id,
+                                             const bool long_size_pre_delay_supported) {
+    command_buffer.GenerateReverbCommand(node_id, effect_info, buffer_offset,
+                                         long_size_pre_delay_supported);
+}
+
+void CommandGenerator::GenerateI3dl2ReverbEffectCommand(const s16 buffer_offset,
+                                                        EffectInfoBase& effect_info,
+                                                        const s32 node_id) {
+    command_buffer.GenerateI3dl2ReverbCommand(node_id, effect_info, buffer_offset);
+}
+
+void CommandGenerator::GenerateAuxCommand(const s16 buffer_offset, EffectInfoBase& effect_info,
+                                          const s32 node_id) {
+
+    if (effect_info.IsEnabled()) {
+        effect_info.GetWorkbuffer(0);
+        effect_info.GetWorkbuffer(1);
+    }
+
+    if (effect_info.GetSendBuffer() != 0 && effect_info.GetReturnBuffer() != 0) {
+        const auto& parameter{
+            *reinterpret_cast<AuxInfo::ParameterVersion1*>(effect_info.GetParameter())};
+        auto channel_index{parameter.mix_buffer_count - 1};
+        u32 write_offset{0};
+        for (u32 i = 0; i < parameter.mix_buffer_count; i++, channel_index--) {
+            auto new_update_count{command_header.sample_count + write_offset};
+            const auto update_count{channel_index > 0 ? 0 : new_update_count};
+            command_buffer.GenerateAuxCommand(node_id, effect_info, parameter.inputs[i],
+                                              parameter.outputs[i], buffer_offset, update_count,
+                                              parameter.count_max, write_offset);
+            write_offset = new_update_count;
+        }
+    }
+}
+
+void CommandGenerator::GenerateBiquadFilterEffectCommand(const s16 buffer_offset,
+                                                         EffectInfoBase& effect_info,
+                                                         const s32 node_id) {
+    const auto& parameter{
+        *reinterpret_cast<BiquadFilterInfo::ParameterVersion1*>(effect_info.GetParameter())};
+    if (effect_info.IsEnabled()) {
+        bool needs_init{false};
+
+        switch (parameter.state) {
+        case EffectInfoBase::ParameterState::Initialized:
+            needs_init = true;
+            break;
+        case EffectInfoBase::ParameterState::Updating:
+        case EffectInfoBase::ParameterState::Updated:
+            if (render_context.behavior->IsBiquadFilterEffectStateClaerBugFixed()) {
+                needs_init = false;
+            } else {
+                needs_init = parameter.state == EffectInfoBase::ParameterState::Updating;
+            }
+            break;
+        default:
+            LOG_ERROR(Service_Audio, "Invalid biquad parameter state {}",
+                      static_cast<u32>(parameter.state));
+            break;
+        }
+
+        for (s8 channel = 0; channel < parameter.channel_count; channel++) {
+            command_buffer.GenerateBiquadFilterCommand(
+                node_id, effect_info, buffer_offset, channel, needs_init,
+                render_context.behavior->UseBiquadFilterFloatProcessing());
+        }
+    } else {
+        for (s8 channel = 0; channel < parameter.channel_count; channel++) {
+            command_buffer.GenerateCopyMixBufferCommand(node_id, effect_info, buffer_offset,
+                                                        channel);
+        }
+    }
+}
+
+void CommandGenerator::GenerateLightLimiterEffectCommand(const s16 buffer_offset,
+                                                         EffectInfoBase& effect_info,
+                                                         const s32 node_id,
+                                                         const u32 effect_index) {
+
+    const auto& state{*reinterpret_cast<LightLimiterInfo::State*>(effect_info.GetStateBuffer())};
+
+    if (render_context.behavior->IsEffectInfoVersion2Supported()) {
+        const auto& parameter{
+            *reinterpret_cast<LightLimiterInfo::ParameterVersion2*>(effect_info.GetParameter())};
+        const auto& result_state{*reinterpret_cast<LightLimiterInfo::StatisticsInternal*>(
+            &effect_context.GetDspSharedResultState(effect_index))};
+        command_buffer.GenerateLightLimiterCommand(node_id, buffer_offset, parameter, result_state,
+                                                   state, effect_info.IsEnabled(),
+                                                   effect_info.GetWorkbuffer(-1));
+    } else {
+        const auto& parameter{
+            *reinterpret_cast<LightLimiterInfo::ParameterVersion1*>(effect_info.GetParameter())};
+        command_buffer.GenerateLightLimiterCommand(node_id, buffer_offset, parameter, state,
+                                                   effect_info.IsEnabled(),
+                                                   effect_info.GetWorkbuffer(-1));
+    }
+}
+
+void CommandGenerator::GenerateCaptureCommand(const s16 buffer_offset, EffectInfoBase& effect_info,
+                                              const s32 node_id) {
+    if (effect_info.IsEnabled()) {
+        effect_info.GetWorkbuffer(0);
+    }
+
+    if (effect_info.GetSendBuffer()) {
+        const auto& parameter{
+            *reinterpret_cast<AuxInfo::ParameterVersion1*>(effect_info.GetParameter())};
+        auto channel_index{parameter.mix_buffer_count - 1};
+        u32 write_offset{0};
+        for (u32 i = 0; i < parameter.mix_buffer_count; i++, channel_index--) {
+            auto new_update_count{command_header.sample_count + write_offset};
+            const auto update_count{channel_index > 0 ? 0 : new_update_count};
+            command_buffer.GenerateCaptureCommand(node_id, effect_info, parameter.inputs[i],
+                                                  parameter.outputs[i], buffer_offset, update_count,
+                                                  parameter.count_max, write_offset);
+            write_offset = new_update_count;
+        }
+    }
+}
+
+void CommandGenerator::GenerateEffectCommand(MixInfo& mix_info) {
+    const auto effect_count{effect_context.GetCount()};
+    for (u32 i = 0; i < effect_count; i++) {
+        const auto effect_index{mix_info.effect_order_buffer[i]};
+        if (effect_index == -1) {
+            break;
+        }
+
+        auto& effect_info = effect_context.GetInfo(effect_index);
+        if (effect_info.ShouldSkip()) {
+            continue;
+        }
+
+        const auto entry_type{mix_info.mix_id == FinalMixId ? PerformanceEntryType::FinalMix
+                                                            : PerformanceEntryType::SubMix};
+
+        switch (effect_info.GetType()) {
+        case EffectInfoBase::Type::Mix: {
+            DetailAspect mix_detail_aspect(*this, entry_type, mix_info.node_id,
+                                           PerformanceDetailType::Unk5);
+            GenerateBufferMixerCommand(mix_info.buffer_offset, effect_info, mix_info.node_id);
+            if (mix_detail_aspect.initialized) {
+                command_buffer.GeneratePerformanceCommand(
+                    mix_detail_aspect.node_id, PerformanceState::Stop,
+                    mix_detail_aspect.performance_entry_address);
+            }
+        } break;
+
+        case EffectInfoBase::Type::Aux: {
+            DetailAspect aux_detail_aspect(*this, entry_type, mix_info.node_id,
+                                           PerformanceDetailType::Unk7);
+            GenerateAuxCommand(mix_info.buffer_offset, effect_info, mix_info.node_id);
+            if (aux_detail_aspect.initialized) {
+                command_buffer.GeneratePerformanceCommand(
+                    aux_detail_aspect.node_id, PerformanceState::Stop,
+                    aux_detail_aspect.performance_entry_address);
+            }
+        } break;
+
+        case EffectInfoBase::Type::Delay: {
+            DetailAspect delay_detail_aspect(*this, entry_type, mix_info.node_id,
+                                             PerformanceDetailType::Unk6);
+            GenerateDelayCommand(mix_info.buffer_offset, effect_info, mix_info.node_id);
+            if (delay_detail_aspect.initialized) {
+                command_buffer.GeneratePerformanceCommand(
+                    delay_detail_aspect.node_id, PerformanceState::Stop,
+                    delay_detail_aspect.performance_entry_address);
+            }
+        } break;
+
+        case EffectInfoBase::Type::Reverb: {
+            DetailAspect reverb_detail_aspect(*this, entry_type, mix_info.node_id,
+                                              PerformanceDetailType::Unk8);
+            GenerateReverbCommand(mix_info.buffer_offset, effect_info, mix_info.node_id,
+                                  render_context.behavior->IsLongSizePreDelaySupported());
+            if (reverb_detail_aspect.initialized) {
+                command_buffer.GeneratePerformanceCommand(
+                    reverb_detail_aspect.node_id, PerformanceState::Stop,
+                    reverb_detail_aspect.performance_entry_address);
+            }
+        } break;
+
+        case EffectInfoBase::Type::I3dl2Reverb: {
+            DetailAspect i3dl2_detail_aspect(*this, entry_type, mix_info.node_id,
+                                             PerformanceDetailType::Unk9);
+            GenerateI3dl2ReverbEffectCommand(mix_info.buffer_offset, effect_info, mix_info.node_id);
+            if (i3dl2_detail_aspect.initialized) {
+                command_buffer.GeneratePerformanceCommand(
+                    i3dl2_detail_aspect.node_id, PerformanceState::Stop,
+                    i3dl2_detail_aspect.performance_entry_address);
+            }
+        } break;
+
+        case EffectInfoBase::Type::BiquadFilter: {
+            DetailAspect biquad_detail_aspect(*this, entry_type, mix_info.node_id,
+                                              PerformanceDetailType::Unk4);
+            GenerateBiquadFilterEffectCommand(mix_info.buffer_offset, effect_info,
+                                              mix_info.node_id);
+            if (biquad_detail_aspect.initialized) {
+                command_buffer.GeneratePerformanceCommand(
+                    biquad_detail_aspect.node_id, PerformanceState::Stop,
+                    biquad_detail_aspect.performance_entry_address);
+            }
+        } break;
+
+        case EffectInfoBase::Type::LightLimiter: {
+            DetailAspect light_limiter_detail_aspect(*this, entry_type, mix_info.node_id,
+                                                     PerformanceDetailType::Unk11);
+            GenerateLightLimiterEffectCommand(mix_info.buffer_offset, effect_info, mix_info.node_id,
+                                              effect_index);
+            if (light_limiter_detail_aspect.initialized) {
+                command_buffer.GeneratePerformanceCommand(
+                    light_limiter_detail_aspect.node_id, PerformanceState::Stop,
+                    light_limiter_detail_aspect.performance_entry_address);
+            }
+        } break;
+
+        case EffectInfoBase::Type::Capture: {
+            DetailAspect capture_detail_aspect(*this, entry_type, mix_info.node_id,
+                                               PerformanceDetailType::Unk12);
+            GenerateCaptureCommand(mix_info.buffer_offset, effect_info, mix_info.node_id);
+            if (capture_detail_aspect.initialized) {
+                command_buffer.GeneratePerformanceCommand(
+                    capture_detail_aspect.node_id, PerformanceState::Stop,
+                    capture_detail_aspect.performance_entry_address);
+            }
+        } break;
+
+        default:
+            LOG_ERROR(Service_Audio, "Invalid effect type {}",
+                      static_cast<u32>(effect_info.GetType()));
+            break;
+        }
+
+        effect_info.UpdateForCommandGeneration();
+    }
+}
+
+void CommandGenerator::GenerateMixCommands(MixInfo& mix_info) {
+    u8 precision{15};
+    if (render_context.behavior->IsVolumeMixParameterPrecisionQ23Supported()) {
+        precision = 23;
+    }
+
+    if (!mix_info.HasAnyConnection()) {
+        return;
+    }
+
+    if (mix_info.dst_mix_id == UnusedMixId) {
+        if (mix_info.dst_splitter_id != UnusedSplitterId) {
+            s16 dest_id{0};
+            auto destination{
+                splitter_context.GetDesintationData(mix_info.dst_splitter_id, dest_id)};
+            while (destination != nullptr) {
+                if (destination->IsConfigured()) {
+                    auto splitter_mix_id{destination->GetMixId()};
+                    if (splitter_mix_id < mix_context.GetCount()) {
+                        auto splitter_mix_info{mix_context.GetInfo(splitter_mix_id)};
+                        const s16 input_index{static_cast<s16>(mix_info.buffer_offset +
+                                                               (dest_id % mix_info.buffer_count))};
+                        for (s16 i = 0; i < splitter_mix_info->buffer_count; i++) {
+                            auto volume{mix_info.volume * destination->GetMixVolume(i)};
+                            if (volume != 0.0f) {
+                                command_buffer.GenerateMixCommand(
+                                    mix_info.node_id, input_index,
+                                    splitter_mix_info->buffer_offset + i, mix_info.buffer_offset,
+                                    volume, precision);
+                            }
+                        }
+                    }
+                }
+                dest_id++;
+                destination =
+                    splitter_context.GetDesintationData(mix_info.dst_splitter_id, dest_id);
+            }
+        }
+    } else {
+        auto dest_mix_info{mix_context.GetInfo(mix_info.dst_mix_id)};
+        for (s16 i = 0; i < mix_info.buffer_count; i++) {
+            for (s16 j = 0; j < dest_mix_info->buffer_count; j++) {
+                auto volume{mix_info.volume * mix_info.mix_volumes[i][j]};
+                if (volume != 0.0f) {
+                    command_buffer.GenerateMixCommand(mix_info.node_id, mix_info.buffer_offset + i,
+                                                      dest_mix_info->buffer_offset + j,
+                                                      mix_info.buffer_offset, volume, precision);
+                }
+            }
+        }
+    }
+}
+
+void CommandGenerator::GenerateSubMixCommand(MixInfo& mix_info) {
+    command_buffer.GenerateDepopForMixBuffersCommand(mix_info.node_id, mix_info,
+                                                     render_context.depop_buffer);
+    GenerateEffectCommand(mix_info);
+
+    DetailAspect mix_detail_aspect(*this, PerformanceEntryType::SubMix, mix_info.node_id,
+                                   PerformanceDetailType::Unk5);
+
+    GenerateMixCommands(mix_info);
+
+    if (mix_detail_aspect.initialized) {
+        command_buffer.GeneratePerformanceCommand(mix_detail_aspect.node_id, PerformanceState::Stop,
+                                                  mix_detail_aspect.performance_entry_address);
+    }
+}
+
+void CommandGenerator::GenerateSubMixCommands() {
+    const auto submix_count{mix_context.GetCount()};
+    for (s32 i = 0; i < submix_count; i++) {
+        auto sorted_info{mix_context.GetSortedInfo(i)};
+        if (!sorted_info->in_use || sorted_info->mix_id == FinalMixId) {
+            continue;
+        }
+
+        EntryAspect submix_entry_aspect(*this, PerformanceEntryType::SubMix, sorted_info->node_id);
+
+        GenerateSubMixCommand(*sorted_info);
+
+        if (submix_entry_aspect.initialized) {
+            command_buffer.GeneratePerformanceCommand(
+                submix_entry_aspect.node_id, PerformanceState::Stop,
+                submix_entry_aspect.performance_entry_address);
+        }
+    }
+}
+
+void CommandGenerator::GenerateFinalMixCommand() {
+    auto& final_mix_info{*mix_context.GetFinalMixInfo()};
+
+    command_buffer.GenerateDepopForMixBuffersCommand(final_mix_info.node_id, final_mix_info,
+                                                     render_context.depop_buffer);
+    GenerateEffectCommand(final_mix_info);
+
+    u8 precision{15};
+    if (render_context.behavior->IsVolumeMixParameterPrecisionQ23Supported()) {
+        precision = 23;
+    }
+
+    for (s16 i = 0; i < final_mix_info.buffer_count; i++) {
+        DetailAspect volume_aspect(*this, PerformanceEntryType::FinalMix, final_mix_info.node_id,
+                                   PerformanceDetailType::Unk3);
+        command_buffer.GenerateVolumeCommand(final_mix_info.node_id, final_mix_info.buffer_offset,
+                                             i, final_mix_info.volume, precision);
+        if (volume_aspect.initialized) {
+            command_buffer.GeneratePerformanceCommand(volume_aspect.node_id, PerformanceState::Stop,
+                                                      volume_aspect.performance_entry_address);
+        }
+    }
+}
+
+void CommandGenerator::GenerateFinalMixCommands() {
+    auto final_mix_info{mix_context.GetFinalMixInfo()};
+    EntryAspect final_mix_entry(*this, PerformanceEntryType::FinalMix, final_mix_info->node_id);
+    GenerateFinalMixCommand();
+    if (final_mix_entry.initialized) {
+        command_buffer.GeneratePerformanceCommand(final_mix_entry.node_id, PerformanceState::Stop,
+                                                  final_mix_entry.performance_entry_address);
+    }
+}
+
+void CommandGenerator::GenerateSinkCommands() {
+    const auto sink_count{sink_context.GetCount()};
+
+    for (u32 i = 0; i < sink_count; i++) {
+        auto sink_info{sink_context.GetInfo(i)};
+        if (sink_info->IsUsed() && sink_info->GetType() == SinkInfoBase::Type::DeviceSink) {
+            auto state{reinterpret_cast<DeviceSinkInfo::DeviceState*>(sink_info->GetState())};
+            if (command_header.sample_rate != TargetSampleRate &&
+                state->upsampler_info == nullptr) {
+                auto device_state{sink_info->GetDeviceState()};
+                device_state->upsampler_info = render_context.upsampler_manager->Allocate();
+            }
+
+            EntryAspect device_sink_entry(*this, PerformanceEntryType::Sink,
+                                          sink_info->GetNodeId());
+            auto final_mix{mix_context.GetFinalMixInfo()};
+            GenerateSinkCommand(final_mix->buffer_offset, *sink_info);
+
+            if (device_sink_entry.initialized) {
+                command_buffer.GeneratePerformanceCommand(
+                    device_sink_entry.node_id, PerformanceState::Stop,
+                    device_sink_entry.performance_entry_address);
+            }
+        }
+    }
+
+    for (u32 i = 0; i < sink_count; i++) {
+        auto sink_info{sink_context.GetInfo(i)};
+        if (sink_info->IsUsed() && sink_info->GetType() == SinkInfoBase::Type::CircularBufferSink) {
+            EntryAspect circular_buffer_entry(*this, PerformanceEntryType::Sink,
+                                              sink_info->GetNodeId());
+            auto final_mix{mix_context.GetFinalMixInfo()};
+            GenerateSinkCommand(final_mix->buffer_offset, *sink_info);
+
+            if (circular_buffer_entry.initialized) {
+                command_buffer.GeneratePerformanceCommand(
+                    circular_buffer_entry.node_id, PerformanceState::Stop,
+                    circular_buffer_entry.performance_entry_address);
+            }
+        }
+    }
+}
+
+void CommandGenerator::GenerateSinkCommand(const s16 buffer_offset, SinkInfoBase& sink_info) {
+    if (sink_info.ShouldSkip()) {
+        return;
+    }
+
+    switch (sink_info.GetType()) {
+    case SinkInfoBase::Type::DeviceSink:
+        GenerateDeviceSinkCommand(buffer_offset, sink_info);
+        break;
+
+    case SinkInfoBase::Type::CircularBufferSink:
+        command_buffer.GenerateCircularBufferSinkCommand(sink_info.GetNodeId(), sink_info,
+                                                         buffer_offset);
+        break;
+
+    default:
+        LOG_ERROR(Service_Audio, "Invalid sink type {}", static_cast<u32>(sink_info.GetType()));
+        break;
+    }
+
+    sink_info.UpdateForCommandGeneration();
+}
+
+void CommandGenerator::GenerateDeviceSinkCommand(const s16 buffer_offset, SinkInfoBase& sink_info) {
+    auto& parameter{
+        *reinterpret_cast<DeviceSinkInfo::DeviceInParameter*>(sink_info.GetParameter())};
+    auto state{*reinterpret_cast<DeviceSinkInfo::DeviceState*>(sink_info.GetState())};
+
+    if (render_context.channels == 2 && parameter.downmix_enabled) {
+        command_buffer.GenerateDownMix6chTo2chCommand(InvalidNodeId, parameter.inputs,
+                                                      buffer_offset, parameter.downmix_coeff);
+    } else if (render_context.channels == 2 && parameter.input_count == 6) {
+        constexpr std::array<f32, 4> default_coeffs{{1.0f, 0.707f, 0.251f, 0.707f}};
+        command_buffer.GenerateDownMix6chTo2chCommand(InvalidNodeId, parameter.inputs,
+                                                      buffer_offset, default_coeffs);
+    }
+
+    if (state.upsampler_info != nullptr) {
+        command_buffer.GenerateUpsampleCommand(
+            InvalidNodeId, buffer_offset, *state.upsampler_info, parameter.input_count,
+            parameter.inputs, command_header.buffer_count, command_header.sample_count,
+            command_header.sample_rate);
+    }
+
+    command_buffer.GenerateDeviceSinkCommand(InvalidNodeId, buffer_offset, sink_info,
+                                             render_context.session_id,
+                                             command_header.samples_buffer);
+}
+
+void CommandGenerator::GeneratePerformanceCommand(
+    s32 node_id, PerformanceState state, const PerformanceEntryAddresses& entry_addresses) {
+    command_buffer.GeneratePerformanceCommand(node_id, state, entry_addresses);
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/command_generator.h

@@ -0,0 +1,339 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <span>
+
+#include "audio_core/renderer/command/commands.h"
+#include "audio_core/renderer/performance/performance_manager.h"
+#include "common/common_types.h"
+
+namespace AudioCore {
+struct AudioRendererSystemContext;
+
+namespace AudioRenderer {
+class CommandBuffer;
+struct CommandListHeader;
+class VoiceContext;
+class MixContext;
+class EffectContext;
+class SplitterContext;
+class SinkContext;
+class BehaviorInfo;
+class VoiceInfo;
+struct VoiceState;
+class MixInfo;
+class SinkInfoBase;
+
+/**
+ * Generates all commands to build up a command list, which are sent to the AudioRender for
+ * processing.
+ */
+class CommandGenerator {
+public:
+    explicit CommandGenerator(CommandBuffer& command_buffer,
+                              const CommandListHeader& command_list_header,
+                              const AudioRendererSystemContext& render_context,
+                              VoiceContext& voice_context, MixContext& mix_context,
+                              EffectContext& effect_context, SinkContext& sink_context,
+                              SplitterContext& splitter_context,
+                              PerformanceManager* performance_manager);
+
+    /**
+     * Calculate the buffer size needed for commands.
+     *
+     * @param behavior - Used to check what features are enabled.
+     * @param params    - Input rendering parameters for numbers of voices/mixes/sinks etc.
+     */
+    static u64 CalculateCommandBufferSize(const BehaviorInfo& behavior,
+                                          const AudioRendererParameterInternal& params) {
+        u64 size{0};
+
+        // Effects
+        size += params.effects * sizeof(EffectInfoBase);
+
+        // Voices
+        u64 voice_size{0};
+        if (behavior.IsWaveBufferVer2Supported()) {
+            voice_size = std::max(std::max(sizeof(AdpcmDataSourceVersion2Command),
+                                           sizeof(PcmInt16DataSourceVersion2Command)),
+                                  sizeof(PcmFloatDataSourceVersion2Command));
+        } else {
+            voice_size = std::max(std::max(sizeof(AdpcmDataSourceVersion1Command),
+                                           sizeof(PcmInt16DataSourceVersion1Command)),
+                                  sizeof(PcmFloatDataSourceVersion1Command));
+        }
+        voice_size += sizeof(BiquadFilterCommand) * MaxBiquadFilters;
+        voice_size += sizeof(VolumeRampCommand);
+        voice_size += sizeof(MixRampGroupedCommand);
+
+        size += params.voices * (params.splitter_infos * sizeof(DepopPrepareCommand) + voice_size);
+
+        // Sub mixes
+        size += sizeof(DepopForMixBuffersCommand) +
+                (sizeof(MixCommand) * MaxMixBuffers) * MaxMixBuffers;
+
+        // Final mix
+        size += sizeof(DepopForMixBuffersCommand) + sizeof(VolumeCommand) * MaxMixBuffers;
+
+        // Splitters
+        size += params.splitter_destinations * sizeof(MixRampCommand) * MaxMixBuffers;
+
+        // Sinks
+        size +=
+            params.sinks * std::max(sizeof(DeviceSinkCommand), sizeof(CircularBufferSinkCommand));
+
+        // Performance
+        size += (params.effects + params.voices + params.sinks + params.sub_mixes + 1 +
+                 PerformanceManager::MaxDetailEntries) *
+                sizeof(PerformanceCommand);
+        return size;
+    }
+
+    /**
+     * Get the current command buffer used to generate commands.
+     *
+     * @return The command buffer.
+     */
+    CommandBuffer& GetCommandBuffer() {
+        return command_buffer;
+    }
+
+    /**
+     * Get the current performance manager,
+     *
+     * @return The performance manager. May be nullptr.
+     */
+    PerformanceManager* GetPerformanceManager() {
+        return performance_manager;
+    }
+
+    /**
+     * Generate a data source command.
+     * These are the basis for all audio output.
+     *
+     * @param voice_info  - Generate the command from this voice.
+     * @param voice_state - State used by the AudioRenderer across calls.
+     * @param channel     - Channel index to generate the command into.
+     */
+    void GenerateDataSourceCommand(VoiceInfo& voice_info, const VoiceState& voice_state,
+                                   s8 channel);
+
+    /**
+     * Generate voice mixing commands.
+     * These are used to mix buffers together, to mix one input to many outputs,
+     * and also used as copy commands to move data around and prevent it being accidentally
+     * overwritten, e.g by another data source command into the same channel.
+     *
+     * @param mix_volumes      - Current volumes of the mix.
+     * @param prev_mix_volumes - Previous volumes of the mix.
+     * @param voice_state      - State used by the AudioRenderer across calls.
+     * @param output_index     - Output mix buffer index.
+     * @param buffer_count     - Number of active mix buffers.
+     * @param input_index      - Input mix buffer index.
+     * @param node_id          - Node id of the voice this command is generated for.
+     */
+    void GenerateVoiceMixCommand(std::span<const f32> mix_volumes,
+                                 std::span<const f32> prev_mix_volumes,
+                                 const VoiceState& voice_state, s16 output_index, s16 buffer_count,
+                                 s16 input_index, s32 node_id);
+
+    /**
+     * Generate a biquad filter command for a voice.
+     *
+     * @param voice_info   - Voice info this command is generated from.
+     * @param voice_state  - State used by the AudioRenderer across calls.
+     * @param buffer_count - Number of active mix buffers.
+     * @param channel      - Channel index of this command.
+     * @param node_id      - Node id of the voice this command is generated for.
+     */
+    void GenerateBiquadFilterCommandForVoice(VoiceInfo& voice_info, const VoiceState& voice_state,
+                                             s16 buffer_count, s8 channel, s32 node_id);
+
+    /**
+     * Generate commands for a voice.
+     * Includes a data source, biquad filter, volume and mixing.
+     *
+     * @param voice_info - Voice info these commands are generated from.
+     */
+    void GenerateVoiceCommand(VoiceInfo& voice_info);
+
+    /**
+     * Generate commands for all voices.
+     */
+    void GenerateVoiceCommands();
+
+    /**
+     * Generate a mixing command.
+     *
+     * @param buffer_offset    - Base mix buffer offset to use.
+     * @param effect_info_base - BufferMixer effect info.
+     * @param node_id          - Node id of the mix this command is generated for.
+     */
+    void GenerateBufferMixerCommand(s16 buffer_offset, EffectInfoBase& effect_info_base,
+                                    s32 node_id);
+
+    /**
+     * Generate a delay effect command.
+     *
+     * @param buffer_offset    - Base mix buffer offset to use.
+     * @param effect_info_base - Delay effect info.
+     * @param node_id          - Node id of the mix this command is generated for.
+     */
+    void GenerateDelayCommand(s16 buffer_offset, EffectInfoBase& effect_info_base, s32 node_id);
+
+    /**
+     * Generate a reverb effect command.
+     *
+     * @param buffer_offset                 - Base mix buffer offset to use.
+     * @param effect_info_base              - Reverb effect info.
+     * @param node_id                       - Node id of the mix this command is generated for.
+     * @param long_size_pre_delay_supported - Use a longer pre-delay time before reverb starts.
+     */
+    void GenerateReverbCommand(s16 buffer_offset, EffectInfoBase& effect_info_base, s32 node_id,
+                               bool long_size_pre_delay_supported);
+
+    /**
+     * Generate an I3DL2 reverb effect command.
+     *
+     * @param buffer_offset    - Base mix buffer offset to use.
+     * @param effect_info_base - I3DL2Reverb effect info.
+     * @param node_id          - Node id of the mix this command is generated for.
+     */
+    void GenerateI3dl2ReverbEffectCommand(s16 buffer_offset, EffectInfoBase& effect_info,
+                                          s32 node_id);
+
+    /**
+     * Generate an aux effect command.
+     *
+     * @param buffer_offset    - Base mix buffer offset to use.
+     * @param effect_info_base - Aux effect info.
+     * @param node_id          - Node id of the mix this command is generated for.
+     */
+    void GenerateAuxCommand(s16 buffer_offset, EffectInfoBase& effect_info, s32 node_id);
+
+    /**
+     * Generate a biquad filter effect command.
+     *
+     * @param buffer_offset    - Base mix buffer offset to use.
+     * @param effect_info_base - Aux effect info.
+     * @param node_id          - Node id of the mix this command is generated for.
+     */
+    void GenerateBiquadFilterEffectCommand(s16 buffer_offset, EffectInfoBase& effect_info,
+                                           s32 node_id);
+
+    /**
+     * Generate a light limiter effect command.
+     *
+     * @param buffer_offset    - Base mix buffer offset to use.
+     * @param effect_info_base - Limiter effect info.
+     * @param node_id          - Node id of the mix this command is generated for.
+     * @param effect_index     - Index for the statistics state.
+     */
+    void GenerateLightLimiterEffectCommand(s16 buffer_offset, EffectInfoBase& effect_info,
+                                           s32 node_id, u32 effect_index);
+
+    /**
+     * Generate a capture effect command.
+     * Writes a mix buffer back to game memory.
+     *
+     * @param buffer_offset    - Base mix buffer offset to use.
+     * @param effect_info_base - Capture effect info.
+     * @param node_id          - Node id of the mix this command is generated for.
+     */
+    void GenerateCaptureCommand(s16 buffer_offset, EffectInfoBase& effect_info, s32 node_id);
+
+    /**
+     * Generate all effect commands for a mix.
+     *
+     * @param mix_info - Mix to generate effects from.
+     */
+    void GenerateEffectCommand(MixInfo& mix_info);
+
+    /**
+     * Generate all mix commands.
+     *
+     * @param mix_info - Mix to generate effects from.
+     */
+    void GenerateMixCommands(MixInfo& mix_info);
+
+    /**
+     * Generate a submix command.
+     * Generates all effects and all mixing commands.
+     *
+     * @param mix_info - Mix to generate effects from.
+     */
+    void GenerateSubMixCommand(MixInfo& mix_info);
+
+    /**
+     * Generate all submix command.
+     */
+    void GenerateSubMixCommands();
+
+    /**
+     * Generate the final mix.
+     */
+    void GenerateFinalMixCommand();
+
+    /**
+     * Generate the final mix commands.
+     */
+    void GenerateFinalMixCommands();
+
+    /**
+     * Generate all sink commands.
+     */
+    void GenerateSinkCommands();
+
+    /**
+     * Generate a sink command.
+     * Sends samples out to the backend, or a game-supplied circular buffer.
+     *
+     * @param buffer_offset - Base mix buffer offset to use.
+     * @param sink_info     - Sink info to generate the commands from.
+     */
+    void GenerateSinkCommand(s16 buffer_offset, SinkInfoBase& sink_info);
+
+    /**
+     * Generate a device sink command.
+     * Sends samples out to the backend.
+     *
+     * @param buffer_offset - Base mix buffer offset to use.
+     * @param sink_info     - Sink info to generate the commands from.
+     */
+    void GenerateDeviceSinkCommand(s16 buffer_offset, SinkInfoBase& sink_info);
+
+    /**
+     * Generate a performance command.
+     * Used to report performance metrics of the AudioRenderer back to the game.
+     *
+     * @param buffer_offset - Base mix buffer offset to use.
+     * @param sink_info     - Sink info to generate the commands from.
+     */
+    void GeneratePerformanceCommand(s32 node_id, PerformanceState state,
+                                    const PerformanceEntryAddresses& entry_addresses);
+
+private:
+    /// Commands will be written by this buffer
+    CommandBuffer& command_buffer;
+    /// Header information for the commands generated
+    const CommandListHeader& command_header;
+    /// Various things to control generation
+    const AudioRendererSystemContext& render_context;
+    /// Used for generating voices
+    VoiceContext& voice_context;
+    /// Used for generating mixes
+    MixContext& mix_context;
+    /// Used for generating effects
+    EffectContext& effect_context;
+    /// Used for generating sinks
+    SinkContext& sink_context;
+    /// Used for generating submixes
+    SplitterContext& splitter_context;
+    /// Used for generating performance
+    PerformanceManager* performance_manager;
+};
+
+} // namespace AudioRenderer
+} // namespace AudioCore

src/audio_core/renderer/command/command_list_header.h

@@ -0,0 +1,22 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <span>
+
+#include "audio_core/common/common.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+
+struct CommandListHeader {
+    u64 buffer_size;
+    u32 command_count;
+    std::span<s32> samples_buffer;
+    s16 buffer_count;
+    u32 sample_count;
+    u32 sample_rate;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/command_processing_time_estimator.h

@@ -0,0 +1,208 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "audio_core/renderer/command/commands.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+/**
+ * Estimate the processing time required for all commands.
+ */
+class ICommandProcessingTimeEstimator {
+public:
+    virtual ~ICommandProcessingTimeEstimator() = default;
+
+    virtual u32 Estimate(const PcmInt16DataSourceVersion1Command& command) const = 0;
+    virtual u32 Estimate(const PcmInt16DataSourceVersion2Command& command) const = 0;
+    virtual u32 Estimate(const PcmFloatDataSourceVersion1Command& command) const = 0;
+    virtual u32 Estimate(const PcmFloatDataSourceVersion2Command& command) const = 0;
+    virtual u32 Estimate(const AdpcmDataSourceVersion1Command& command) const = 0;
+    virtual u32 Estimate(const AdpcmDataSourceVersion2Command& command) const = 0;
+    virtual u32 Estimate(const VolumeCommand& command) const = 0;
+    virtual u32 Estimate(const VolumeRampCommand& command) const = 0;
+    virtual u32 Estimate(const BiquadFilterCommand& command) const = 0;
+    virtual u32 Estimate(const MixCommand& command) const = 0;
+    virtual u32 Estimate(const MixRampCommand& command) const = 0;
+    virtual u32 Estimate(const MixRampGroupedCommand& command) const = 0;
+    virtual u32 Estimate(const DepopPrepareCommand& command) const = 0;
+    virtual u32 Estimate(const DepopForMixBuffersCommand& command) const = 0;
+    virtual u32 Estimate(const DelayCommand& command) const = 0;
+    virtual u32 Estimate(const UpsampleCommand& command) const = 0;
+    virtual u32 Estimate(const DownMix6chTo2chCommand& command) const = 0;
+    virtual u32 Estimate(const AuxCommand& command) const = 0;
+    virtual u32 Estimate(const DeviceSinkCommand& command) const = 0;
+    virtual u32 Estimate(const CircularBufferSinkCommand& command) const = 0;
+    virtual u32 Estimate(const ReverbCommand& command) const = 0;
+    virtual u32 Estimate(const I3dl2ReverbCommand& command) const = 0;
+    virtual u32 Estimate(const PerformanceCommand& command) const = 0;
+    virtual u32 Estimate(const ClearMixBufferCommand& command) const = 0;
+    virtual u32 Estimate(const CopyMixBufferCommand& command) const = 0;
+    virtual u32 Estimate(const LightLimiterVersion1Command& command) const = 0;
+    virtual u32 Estimate(const LightLimiterVersion2Command& command) const = 0;
+    virtual u32 Estimate(const MultiTapBiquadFilterCommand& command) const = 0;
+    virtual u32 Estimate(const CaptureCommand& command) const = 0;
+};
+
+class CommandProcessingTimeEstimatorVersion1 final : public ICommandProcessingTimeEstimator {
+public:
+    CommandProcessingTimeEstimatorVersion1(u32 sample_count_, u32 buffer_count_)
+        : sample_count{sample_count_}, buffer_count{buffer_count_} {}
+
+    u32 Estimate(const PcmInt16DataSourceVersion1Command& command) const override;
+    u32 Estimate(const PcmInt16DataSourceVersion2Command& command) const override;
+    u32 Estimate(const PcmFloatDataSourceVersion1Command& command) const override;
+    u32 Estimate(const PcmFloatDataSourceVersion2Command& command) const override;
+    u32 Estimate(const AdpcmDataSourceVersion1Command& command) const override;
+    u32 Estimate(const AdpcmDataSourceVersion2Command& command) const override;
+    u32 Estimate(const VolumeCommand& command) const override;
+    u32 Estimate(const VolumeRampCommand& command) const override;
+    u32 Estimate(const BiquadFilterCommand& command) const override;
+    u32 Estimate(const MixCommand& command) const override;
+    u32 Estimate(const MixRampCommand& command) const override;
+    u32 Estimate(const MixRampGroupedCommand& command) const override;
+    u32 Estimate(const DepopPrepareCommand& command) const override;
+    u32 Estimate(const DepopForMixBuffersCommand& command) const override;
+    u32 Estimate(const DelayCommand& command) const override;
+    u32 Estimate(const UpsampleCommand& command) const override;
+    u32 Estimate(const DownMix6chTo2chCommand& command) const override;
+    u32 Estimate(const AuxCommand& command) const override;
+    u32 Estimate(const DeviceSinkCommand& command) const override;
+    u32 Estimate(const CircularBufferSinkCommand& command) const override;
+    u32 Estimate(const ReverbCommand& command) const override;
+    u32 Estimate(const I3dl2ReverbCommand& command) const override;
+    u32 Estimate(const PerformanceCommand& command) const override;
+    u32 Estimate(const ClearMixBufferCommand& command) const override;
+    u32 Estimate(const CopyMixBufferCommand& command) const override;
+    u32 Estimate(const LightLimiterVersion1Command& command) const override;
+    u32 Estimate(const LightLimiterVersion2Command& command) const override;
+    u32 Estimate(const MultiTapBiquadFilterCommand& command) const override;
+    u32 Estimate(const CaptureCommand& command) const override;
+
+private:
+    u32 sample_count{};
+    u32 buffer_count{};
+};
+
+class CommandProcessingTimeEstimatorVersion2 final : public ICommandProcessingTimeEstimator {
+public:
+    CommandProcessingTimeEstimatorVersion2(u32 sample_count_, u32 buffer_count_)
+        : sample_count{sample_count_}, buffer_count{buffer_count_} {}
+
+    u32 Estimate(const PcmInt16DataSourceVersion1Command& command) const override;
+    u32 Estimate(const PcmInt16DataSourceVersion2Command& command) const override;
+    u32 Estimate(const PcmFloatDataSourceVersion1Command& command) const override;
+    u32 Estimate(const PcmFloatDataSourceVersion2Command& command) const override;
+    u32 Estimate(const AdpcmDataSourceVersion1Command& command) const override;
+    u32 Estimate(const AdpcmDataSourceVersion2Command& command) const override;
+    u32 Estimate(const VolumeCommand& command) const override;
+    u32 Estimate(const VolumeRampCommand& command) const override;
+    u32 Estimate(const BiquadFilterCommand& command) const override;
+    u32 Estimate(const MixCommand& command) const override;
+    u32 Estimate(const MixRampCommand& command) const override;
+    u32 Estimate(const MixRampGroupedCommand& command) const override;
+    u32 Estimate(const DepopPrepareCommand& command) const override;
+    u32 Estimate(const DepopForMixBuffersCommand& command) const override;
+    u32 Estimate(const DelayCommand& command) const override;
+    u32 Estimate(const UpsampleCommand& command) const override;
+    u32 Estimate(const DownMix6chTo2chCommand& command) const override;
+    u32 Estimate(const AuxCommand& command) const override;
+    u32 Estimate(const DeviceSinkCommand& command) const override;
+    u32 Estimate(const CircularBufferSinkCommand& command) const override;
+    u32 Estimate(const ReverbCommand& command) const override;
+    u32 Estimate(const I3dl2ReverbCommand& command) const override;
+    u32 Estimate(const PerformanceCommand& command) const override;
+    u32 Estimate(const ClearMixBufferCommand& command) const override;
+    u32 Estimate(const CopyMixBufferCommand& command) const override;
+    u32 Estimate(const LightLimiterVersion1Command& command) const override;
+    u32 Estimate(const LightLimiterVersion2Command& command) const override;
+    u32 Estimate(const MultiTapBiquadFilterCommand& command) const override;
+    u32 Estimate(const CaptureCommand& command) const override;
+
+private:
+    u32 sample_count{};
+    u32 buffer_count{};
+};
+
+class CommandProcessingTimeEstimatorVersion3 final : public ICommandProcessingTimeEstimator {
+public:
+    CommandProcessingTimeEstimatorVersion3(u32 sample_count_, u32 buffer_count_)
+        : sample_count{sample_count_}, buffer_count{buffer_count_} {}
+
+    u32 Estimate(const PcmInt16DataSourceVersion1Command& command) const override;
+    u32 Estimate(const PcmInt16DataSourceVersion2Command& command) const override;
+    u32 Estimate(const PcmFloatDataSourceVersion1Command& command) const override;
+    u32 Estimate(const PcmFloatDataSourceVersion2Command& command) const override;
+    u32 Estimate(const AdpcmDataSourceVersion1Command& command) const override;
+    u32 Estimate(const AdpcmDataSourceVersion2Command& command) const override;
+    u32 Estimate(const VolumeCommand& command) const override;
+    u32 Estimate(const VolumeRampCommand& command) const override;
+    u32 Estimate(const BiquadFilterCommand& command) const override;
+    u32 Estimate(const MixCommand& command) const override;
+    u32 Estimate(const MixRampCommand& command) const override;
+    u32 Estimate(const MixRampGroupedCommand& command) const override;
+    u32 Estimate(const DepopPrepareCommand& command) const override;
+    u32 Estimate(const DepopForMixBuffersCommand& command) const override;
+    u32 Estimate(const DelayCommand& command) const override;
+    u32 Estimate(const UpsampleCommand& command) const override;
+    u32 Estimate(const DownMix6chTo2chCommand& command) const override;
+    u32 Estimate(const AuxCommand& command) const override;
+    u32 Estimate(const DeviceSinkCommand& command) const override;
+    u32 Estimate(const CircularBufferSinkCommand& command) const override;
+    u32 Estimate(const ReverbCommand& command) const override;
+    u32 Estimate(const I3dl2ReverbCommand& command) const override;
+    u32 Estimate(const PerformanceCommand& command) const override;
+    u32 Estimate(const ClearMixBufferCommand& command) const override;
+    u32 Estimate(const CopyMixBufferCommand& command) const override;
+    u32 Estimate(const LightLimiterVersion1Command& command) const override;
+    u32 Estimate(const LightLimiterVersion2Command& command) const override;
+    u32 Estimate(const MultiTapBiquadFilterCommand& command) const override;
+    u32 Estimate(const CaptureCommand& command) const override;
+
+private:
+    u32 sample_count{};
+    u32 buffer_count{};
+};
+
+class CommandProcessingTimeEstimatorVersion4 final : public ICommandProcessingTimeEstimator {
+public:
+    CommandProcessingTimeEstimatorVersion4(u32 sample_count_, u32 buffer_count_)
+        : sample_count{sample_count_}, buffer_count{buffer_count_} {}
+
+    u32 Estimate(const PcmInt16DataSourceVersion1Command& command) const override;
+    u32 Estimate(const PcmInt16DataSourceVersion2Command& command) const override;
+    u32 Estimate(const PcmFloatDataSourceVersion1Command& command) const override;
+    u32 Estimate(const PcmFloatDataSourceVersion2Command& command) const override;
+    u32 Estimate(const AdpcmDataSourceVersion1Command& command) const override;
+    u32 Estimate(const AdpcmDataSourceVersion2Command& command) const override;
+    u32 Estimate(const VolumeCommand& command) const override;
+    u32 Estimate(const VolumeRampCommand& command) const override;
+    u32 Estimate(const BiquadFilterCommand& command) const override;
+    u32 Estimate(const MixCommand& command) const override;
+    u32 Estimate(const MixRampCommand& command) const override;
+    u32 Estimate(const MixRampGroupedCommand& command) const override;
+    u32 Estimate(const DepopPrepareCommand& command) const override;
+    u32 Estimate(const DepopForMixBuffersCommand& command) const override;
+    u32 Estimate(const DelayCommand& command) const override;
+    u32 Estimate(const UpsampleCommand& command) const override;
+    u32 Estimate(const DownMix6chTo2chCommand& command) const override;
+    u32 Estimate(const AuxCommand& command) const override;
+    u32 Estimate(const DeviceSinkCommand& command) const override;
+    u32 Estimate(const CircularBufferSinkCommand& command) const override;
+    u32 Estimate(const ReverbCommand& command) const override;
+    u32 Estimate(const I3dl2ReverbCommand& command) const override;
+    u32 Estimate(const PerformanceCommand& command) const override;
+    u32 Estimate(const ClearMixBufferCommand& command) const override;
+    u32 Estimate(const CopyMixBufferCommand& command) const override;
+    u32 Estimate(const LightLimiterVersion1Command& command) const override;
+    u32 Estimate(const LightLimiterVersion2Command& command) const override;
+    u32 Estimate(const MultiTapBiquadFilterCommand& command) const override;
+    u32 Estimate(const CaptureCommand& command) const override;
+
+private:
+    u32 sample_count{};
+    u32 buffer_count{};
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/commands.h

@@ -0,0 +1,31 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "audio_core/renderer/command/data_source/adpcm.h"
+#include "audio_core/renderer/command/data_source/pcm_float.h"
+#include "audio_core/renderer/command/data_source/pcm_int16.h"
+#include "audio_core/renderer/command/effect/aux_.h"
+#include "audio_core/renderer/command/effect/biquad_filter.h"
+#include "audio_core/renderer/command/effect/capture.h"
+#include "audio_core/renderer/command/effect/delay.h"
+#include "audio_core/renderer/command/effect/i3dl2_reverb.h"
+#include "audio_core/renderer/command/effect/light_limiter.h"
+#include "audio_core/renderer/command/effect/multi_tap_biquad_filter.h"
+#include "audio_core/renderer/command/effect/reverb.h"
+#include "audio_core/renderer/command/icommand.h"
+#include "audio_core/renderer/command/mix/clear_mix.h"
+#include "audio_core/renderer/command/mix/copy_mix.h"
+#include "audio_core/renderer/command/mix/depop_for_mix_buffers.h"
+#include "audio_core/renderer/command/mix/depop_prepare.h"
+#include "audio_core/renderer/command/mix/mix.h"
+#include "audio_core/renderer/command/mix/mix_ramp.h"
+#include "audio_core/renderer/command/mix/mix_ramp_grouped.h"
+#include "audio_core/renderer/command/mix/volume.h"
+#include "audio_core/renderer/command/mix/volume_ramp.h"
+#include "audio_core/renderer/command/performance/performance.h"
+#include "audio_core/renderer/command/resample/downmix_6ch_to_2ch.h"
+#include "audio_core/renderer/command/resample/upsample.h"
+#include "audio_core/renderer/command/sink/circular_buffer.h"
+#include "audio_core/renderer/command/sink/device.h"

src/audio_core/renderer/command/data_source/adpcm.cpp

@@ -0,0 +1,84 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <span>
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/data_source/adpcm.h"
+#include "audio_core/renderer/command/data_source/decode.h"
+
+namespace AudioCore::AudioRenderer {
+
+void AdpcmDataSourceVersion1Command::Dump(const ADSP::CommandListProcessor& processor,
+                                          std::string& string) {
+    string += fmt::format("AdpcmDataSourceVersion1Command\n\toutput_index {:02X} source sample "
+                          "rate {} target sample rate {} src quality {}\n",
+                          output_index, sample_rate, processor.target_sample_rate, src_quality);
+}
+
+void AdpcmDataSourceVersion1Command::Process(const ADSP::CommandListProcessor& processor) {
+    auto out_buffer{processor.mix_buffers.subspan(output_index * processor.sample_count,
+                                                  processor.sample_count)};
+
+    DecodeFromWaveBuffersArgs args{
+        .sample_format{SampleFormat::Adpcm},
+        .output{out_buffer},
+        .voice_state{reinterpret_cast<VoiceState*>(voice_state)},
+        .wave_buffers{wave_buffers},
+        .channel{0},
+        .channel_count{1},
+        .src_quality{src_quality},
+        .pitch{pitch},
+        .source_sample_rate{sample_rate},
+        .target_sample_rate{processor.target_sample_rate},
+        .sample_count{processor.sample_count},
+        .data_address{data_address},
+        .data_size{data_size},
+        .IsVoicePlayedSampleCountResetAtLoopPointSupported{(flags & 1) != 0},
+        .IsVoicePitchAndSrcSkippedSupported{(flags & 2) != 0},
+    };
+
+    DecodeFromWaveBuffers(*processor.memory, args);
+}
+
+bool AdpcmDataSourceVersion1Command::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+void AdpcmDataSourceVersion2Command::Dump(const ADSP::CommandListProcessor& processor,
+                                          std::string& string) {
+    string += fmt::format("AdpcmDataSourceVersion2Command\n\toutput_index {:02X} source sample "
+                          "rate {} target sample rate {} src quality {}\n",
+                          output_index, sample_rate, processor.target_sample_rate, src_quality);
+}
+
+void AdpcmDataSourceVersion2Command::Process(const ADSP::CommandListProcessor& processor) {
+    auto out_buffer{processor.mix_buffers.subspan(output_index * processor.sample_count,
+                                                  processor.sample_count)};
+
+    DecodeFromWaveBuffersArgs args{
+        .sample_format{SampleFormat::Adpcm},
+        .output{out_buffer},
+        .voice_state{reinterpret_cast<VoiceState*>(voice_state)},
+        .wave_buffers{wave_buffers},
+        .channel{0},
+        .channel_count{1},
+        .src_quality{src_quality},
+        .pitch{pitch},
+        .source_sample_rate{sample_rate},
+        .target_sample_rate{processor.target_sample_rate},
+        .sample_count{processor.sample_count},
+        .data_address{data_address},
+        .data_size{data_size},
+        .IsVoicePlayedSampleCountResetAtLoopPointSupported{(flags & 1) != 0},
+        .IsVoicePitchAndSrcSkippedSupported{(flags & 2) != 0},
+    };
+
+    DecodeFromWaveBuffers(*processor.memory, args);
+}
+
+bool AdpcmDataSourceVersion2Command::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/data_source/adpcm.h

@@ -0,0 +1,119 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <string>
+
+#include "audio_core/common/common.h"
+#include "audio_core/common/wave_buffer.h"
+#include "audio_core/renderer/command/icommand.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command to decode ADPCM-encoded version 1 wavebuffers
+ * into the output_index mix buffer.
+ */
+struct AdpcmDataSourceVersion1Command : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Quality used for sample rate conversion
+    SrcQuality src_quality;
+    /// Mix buffer index for decoded samples
+    s16 output_index;
+    /// Flags to control decoding (see AudioCore::AudioRenderer::VoiceInfo::Flags)
+    u16 flags;
+    /// Wavebuffer sample rate
+    u32 sample_rate;
+    /// Pitch used for sample rate conversion
+    f32 pitch;
+    /// Wavebuffers containing the wavebuffer address, context address, looping information etc
+    std::array<WaveBufferVersion2, MaxWaveBuffers> wave_buffers;
+    /// Voice state, updated each call and written back to game
+    CpuAddr voice_state;
+    /// Coefficients data address
+    CpuAddr data_address;
+    /// Coefficients data size
+    u64 data_size;
+};
+
+/**
+ * AudioRenderer command to decode ADPCM-encoded version 2 wavebuffers
+ * into the output_index mix buffer.
+ */
+struct AdpcmDataSourceVersion2Command : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Quality used for sample rate conversion
+    SrcQuality src_quality;
+    /// Mix buffer index for decoded samples
+    s16 output_index;
+    /// Flags to control decoding (see AudioCore::AudioRenderer::VoiceInfo::Flags)
+    u16 flags;
+    /// Wavebuffer sample rate
+    u32 sample_rate;
+    /// Pitch used for sample rate conversion
+    f32 pitch;
+    /// Target channel to read within the wavebuffer
+    s8 channel_index;
+    /// Number of channels within the wavebuffer
+    s8 channel_count;
+    /// Wavebuffers containing the wavebuffer address, context address, looping information etc
+    std::array<WaveBufferVersion2, MaxWaveBuffers> wave_buffers;
+    /// Voice state, updated each call and written back to game
+    CpuAddr voice_state;
+    /// Coefficients data address
+    CpuAddr data_address;
+    /// Coefficients data size
+    u64 data_size;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/data_source/decode.cpp

@@ -0,0 +1,428 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <array>
+#include <vector>
+
+#include "audio_core/renderer/command/data_source/decode.h"
+#include "audio_core/renderer/command/resample/resample.h"
+#include "common/fixed_point.h"
+#include "common/logging/log.h"
+#include "core/memory.h"
+
+namespace AudioCore::AudioRenderer {
+
+constexpr u32 TempBufferSize = 0x3F00;
+constexpr std::array<u8, 3> PitchBySrcQuality = {4, 8, 4};
+
+/**
+ * Decode PCM data. Only s16 or f32 is supported.
+ *
+ * @tparam T         - Type to decode. Only s16 and f32 are supported.
+ * @param memory     - Core memory for reading samples.
+ * @param out_buffer - Output mix buffer to receive the samples.
+ * @param req        - Information for how to decode.
+ * @return Number of samples decoded.
+ */
+template <typename T>
+static u32 DecodePcm(Core::Memory::Memory& memory, std::span<s16> out_buffer,
+                     const DecodeArg& req) {
+    constexpr s32 min{std::numeric_limits<s16>::min()};
+    constexpr s32 max{std::numeric_limits<s16>::max()};
+
+    if (req.buffer == 0 || req.buffer_size == 0) {
+        return 0;
+    }
+
+    if (req.start_offset >= req.end_offset) {
+        return 0;
+    }
+
+    auto samples_to_decode{
+        std::min(req.samples_to_read, req.end_offset - req.start_offset - req.offset)};
+    u32 channel_count{static_cast<u32>(req.channel_count)};
+
+    switch (req.channel_count) {
+    default: {
+        const VAddr source{req.buffer +
+                           (((req.start_offset + req.offset) * channel_count) * sizeof(T))};
+        const u64 size{channel_count * samples_to_decode};
+        const u64 size_bytes{size * sizeof(T)};
+
+        std::vector<T> samples(size);
+        memory.ReadBlockUnsafe(source, samples.data(), size_bytes);
+
+        if constexpr (std::is_floating_point_v<T>) {
+            for (u32 i = 0; i < samples_to_decode; i++) {
+                auto sample{static_cast<s32>(samples[i * channel_count + req.target_channel] *
+                                             std::numeric_limits<s16>::max())};
+                out_buffer[i] = static_cast<s16>(std::clamp(sample, min, max));
+            }
+        } else {
+            for (u32 i = 0; i < samples_to_decode; i++) {
+                out_buffer[i] = samples[i * channel_count + req.target_channel];
+            }
+        }
+    } break;
+
+    case 1:
+        if (req.target_channel != 0) {
+            LOG_ERROR(Service_Audio, "Invalid target channel, expected 0, got {}",
+                      req.target_channel);
+            return 0;
+        }
+
+        const VAddr source{req.buffer + ((req.start_offset + req.offset) * sizeof(T))};
+        std::vector<T> samples(samples_to_decode);
+        memory.ReadBlockUnsafe(source, samples.data(), samples_to_decode * sizeof(T));
+
+        if constexpr (std::is_floating_point_v<T>) {
+            for (u32 i = 0; i < samples_to_decode; i++) {
+                auto sample{static_cast<s32>(samples[i * channel_count + req.target_channel] *
+                                             std::numeric_limits<s16>::max())};
+                out_buffer[i] = static_cast<s16>(std::clamp(sample, min, max));
+            }
+        } else {
+            std::memcpy(out_buffer.data(), samples.data(), samples_to_decode * sizeof(s16));
+        }
+        break;
+    }
+
+    return samples_to_decode;
+}
+
+/**
+ * Decode ADPCM data.
+ *
+ * @param memory     - Core memory for reading samples.
+ * @param out_buffer - Output mix buffer to receive the samples.
+ * @param req        - Information for how to decode.
+ * @return Number of samples decoded.
+ */
+static u32 DecodeAdpcm(Core::Memory::Memory& memory, std::span<s16> out_buffer,
+                       const DecodeArg& req) {
+    constexpr u32 SamplesPerFrame{14};
+    constexpr u32 NibblesPerFrame{16};
+
+    if (req.buffer == 0 || req.buffer_size == 0) {
+        LOG_ERROR(Service_Audio, "Buffer is 0!");
+        return 0;
+    }
+
+    if (req.start_offset >= req.end_offset) {
+        LOG_ERROR(Service_Audio, "Start offset greater than end offset!");
+        return 0;
+    }
+
+    auto end{(req.end_offset % SamplesPerFrame) +
+             NibblesPerFrame * (req.end_offset / SamplesPerFrame)};
+    if (req.end_offset % SamplesPerFrame) {
+        end += 3;
+    } else {
+        end += 1;
+    }
+
+    if (end / 2 > req.buffer_size) {
+        LOG_ERROR(Service_Audio, "End greater than buffer size!");
+        return 0;
+    }
+
+    auto samples_to_process{
+        std::min(req.end_offset - req.start_offset - req.offset, req.samples_to_read)};
+
+    auto samples_to_read{samples_to_process};
+    auto start_pos{req.start_offset + req.offset};
+    auto samples_remaining_in_frame{start_pos % SamplesPerFrame};
+    auto position_in_frame{(start_pos / SamplesPerFrame) * NibblesPerFrame +
+                           samples_remaining_in_frame};
+
+    if (samples_remaining_in_frame) {
+        position_in_frame += 2;
+    }
+
+    const auto size{std::max((samples_to_process / 8U) * SamplesPerFrame, 8U)};
+    std::vector<u8> wavebuffer(size, 0);
+    memory.ReadBlockUnsafe(req.buffer + position_in_frame / 2, wavebuffer.data(),
+                           wavebuffer.size());
+
+    auto context{req.adpcm_context};
+    auto header{context->header};
+    u8 coeff_index{static_cast<u8>((header >> 4U) & 0xFU)};
+    u8 scale{static_cast<u8>(header & 0xFU)};
+    s32 coeff0{req.coefficients[coeff_index * 2 + 0]};
+    s32 coeff1{req.coefficients[coeff_index * 2 + 1]};
+
+    auto yn0{context->yn0};
+    auto yn1{context->yn1};
+
+    static constexpr std::array<s32, 16> Steps{
+        0, 1, 2, 3, 4, 5, 6, 7, -8, -7, -6, -5, -4, -3, -2, -1,
+    };
+
+    const auto decode_sample = [&](const s32 code) -> s16 {
+        const auto xn = code * (1 << scale);
+        const auto prediction = coeff0 * yn0 + coeff1 * yn1;
+        const auto sample = ((xn << 11) + 0x400 + prediction) >> 11;
+        const auto saturated = std::clamp<s32>(sample, -0x8000, 0x7FFF);
+        yn1 = yn0;
+        yn0 = static_cast<s16>(saturated);
+        return yn0;
+    };
+
+    u32 read_index{0};
+    u32 write_index{0};
+
+    while (samples_to_read > 0) {
+        // Are we at a new frame?
+        if ((position_in_frame % NibblesPerFrame) == 0) {
+            header = wavebuffer[read_index++];
+            coeff_index = (header >> 4) & 0xF;
+            scale = header & 0xF;
+            coeff0 = req.coefficients[coeff_index * 2 + 0];
+            coeff1 = req.coefficients[coeff_index * 2 + 1];
+            position_in_frame += 2;
+
+            // Can we consume all of this frame's samples?
+            if (samples_to_read >= SamplesPerFrame) {
+                // Can grab all samples until the next header
+                for (u32 i = 0; i < SamplesPerFrame / 2; i++) {
+                    auto code0{Steps[(wavebuffer[read_index] >> 4) & 0xF]};
+                    auto code1{Steps[wavebuffer[read_index] & 0xF]};
+                    read_index++;
+
+                    out_buffer[write_index++] = decode_sample(code0);
+                    out_buffer[write_index++] = decode_sample(code1);
+                }
+
+                position_in_frame += SamplesPerFrame;
+                samples_to_read -= SamplesPerFrame;
+                continue;
+            }
+        }
+
+        // Decode a single sample
+        auto code{wavebuffer[read_index]};
+        if (position_in_frame & 1) {
+            code &= 0xF;
+            read_index++;
+        } else {
+            code >>= 4;
+        }
+
+        out_buffer[write_index++] = decode_sample(Steps[code]);
+
+        position_in_frame++;
+        samples_to_read--;
+    }
+
+    context->header = header;
+    context->yn0 = yn0;
+    context->yn1 = yn1;
+
+    return samples_to_process;
+}
+
+/**
+ * Decode implementation.
+ * Decode wavebuffers according to the given args.
+ *
+ * @param memory - Core memory to read data from.
+ * @param args   - The wavebuffer data, and information for how to decode it.
+ */
+void DecodeFromWaveBuffers(Core::Memory::Memory& memory, const DecodeFromWaveBuffersArgs& args) {
+    auto& voice_state{*args.voice_state};
+    auto remaining_sample_count{args.sample_count};
+    auto fraction{voice_state.fraction};
+
+    const auto sample_rate_ratio{
+        (Common::FixedPoint<49, 15>(args.source_sample_rate) / args.target_sample_rate) *
+        args.pitch};
+    const auto size_required{fraction + remaining_sample_count * sample_rate_ratio};
+
+    if (size_required.to_int_floor() < 0) {
+        return;
+    }
+
+    auto pitch{PitchBySrcQuality[static_cast<u32>(args.src_quality)]};
+    if (static_cast<u32>(pitch + size_required.to_int_floor()) > TempBufferSize) {
+        return;
+    }
+
+    auto max_remaining_sample_count{
+        ((Common::FixedPoint<17, 15>(TempBufferSize) - fraction) / sample_rate_ratio)
+            .to_uint_floor()};
+    max_remaining_sample_count = std::min(max_remaining_sample_count, remaining_sample_count);
+
+    auto wavebuffers_consumed{voice_state.wave_buffers_consumed};
+    auto wavebuffer_index{voice_state.wave_buffer_index};
+    auto played_sample_count{voice_state.played_sample_count};
+
+    bool is_buffer_starved{false};
+    u32 offset{voice_state.offset};
+
+    auto output_buffer{args.output};
+    std::vector<s16> temp_buffer(TempBufferSize, 0);
+
+    while (remaining_sample_count > 0) {
+        const auto samples_to_write{std::min(remaining_sample_count, max_remaining_sample_count)};
+        const auto samples_to_read{
+            (fraction + samples_to_write * sample_rate_ratio).to_uint_floor()};
+
+        u32 temp_buffer_pos{0};
+
+        if (!args.IsVoicePitchAndSrcSkippedSupported) {
+            for (u32 i = 0; i < pitch; i++) {
+                temp_buffer[i] = voice_state.sample_history[i];
+            }
+            temp_buffer_pos = pitch;
+        }
+
+        u32 samples_read{0};
+        while (samples_read < samples_to_read) {
+            if (wavebuffer_index >= MaxWaveBuffers) {
+                LOG_ERROR(Service_Audio, "Invalid wavebuffer index! {}", wavebuffer_index);
+                wavebuffer_index = 0;
+                voice_state.wave_buffer_valid.fill(false);
+                wavebuffers_consumed = MaxWaveBuffers;
+            }
+
+            if (!voice_state.wave_buffer_valid[wavebuffer_index]) {
+                is_buffer_starved = true;
+                break;
+            }
+
+            auto& wavebuffer{args.wave_buffers[wavebuffer_index]};
+
+            if (offset == 0 && args.sample_format == SampleFormat::Adpcm &&
+                wavebuffer.context != 0) {
+                memory.ReadBlockUnsafe(wavebuffer.context, &voice_state.adpcm_context,
+                                       wavebuffer.context_size);
+            }
+
+            auto start_offset{wavebuffer.start_offset};
+            auto end_offset{wavebuffer.end_offset};
+
+            if (wavebuffer.loop && voice_state.loop_count > 0 &&
+                wavebuffer.loop_start_offset != 0 && wavebuffer.loop_end_offset != 0 &&
+                wavebuffer.loop_start_offset <= wavebuffer.loop_end_offset) {
+                start_offset = wavebuffer.loop_start_offset;
+                end_offset = wavebuffer.loop_end_offset;
+            }
+
+            DecodeArg decode_arg{.buffer{wavebuffer.buffer},
+                                 .buffer_size{wavebuffer.buffer_size},
+                                 .start_offset{start_offset},
+                                 .end_offset{end_offset},
+                                 .channel_count{args.channel_count},
+                                 .coefficients{},
+                                 .adpcm_context{nullptr},
+                                 .target_channel{args.channel},
+                                 .offset{offset},
+                                 .samples_to_read{samples_to_read - samples_read}};
+
+            s32 samples_decoded{0};
+
+            switch (args.sample_format) {
+            case SampleFormat::PcmInt16:
+                samples_decoded = DecodePcm<s16>(
+                    memory, {&temp_buffer[temp_buffer_pos], args.sample_count}, decode_arg);
+                break;
+
+            case SampleFormat::PcmFloat:
+                samples_decoded = DecodePcm<f32>(
+                    memory, {&temp_buffer[temp_buffer_pos], args.sample_count}, decode_arg);
+                break;
+
+            case SampleFormat::Adpcm: {
+                decode_arg.adpcm_context = &voice_state.adpcm_context;
+                memory.ReadBlockUnsafe(args.data_address, &decode_arg.coefficients, args.data_size);
+                samples_decoded = DecodeAdpcm(
+                    memory, {&temp_buffer[temp_buffer_pos], args.sample_count}, decode_arg);
+            } break;
+
+            default:
+                LOG_ERROR(Service_Audio, "Invalid sample format to decode {}",
+                          static_cast<u32>(args.sample_format));
+                samples_decoded = 0;
+                break;
+            }
+
+            played_sample_count += samples_decoded;
+            samples_read += samples_decoded;
+            temp_buffer_pos += samples_decoded;
+            offset += samples_decoded;
+
+            if (samples_decoded == 0 || offset >= end_offset - start_offset) {
+                offset = 0;
+                if (!wavebuffer.loop) {
+                    voice_state.wave_buffer_valid[wavebuffer_index] = false;
+                    voice_state.loop_count = 0;
+
+                    if (wavebuffer.stream_ended) {
+                        played_sample_count = 0;
+                    }
+
+                    wavebuffer_index = (wavebuffer_index + 1) % MaxWaveBuffers;
+                    wavebuffers_consumed++;
+                } else {
+                    voice_state.loop_count++;
+                    if (wavebuffer.loop_count > 0 &&
+                        (voice_state.loop_count > wavebuffer.loop_count || samples_decoded == 0)) {
+                        voice_state.wave_buffer_valid[wavebuffer_index] = false;
+                        voice_state.loop_count = 0;
+
+                        if (wavebuffer.stream_ended) {
+                            played_sample_count = 0;
+                        }
+
+                        wavebuffer_index = (wavebuffer_index + 1) % MaxWaveBuffers;
+                        wavebuffers_consumed++;
+                    }
+
+                    if (samples_decoded == 0) {
+                        is_buffer_starved = true;
+                        break;
+                    }
+
+                    if (args.IsVoicePlayedSampleCountResetAtLoopPointSupported) {
+                        played_sample_count = 0;
+                    }
+                }
+            }
+        }
+
+        if (args.IsVoicePitchAndSrcSkippedSupported) {
+            if (samples_read > output_buffer.size()) {
+                LOG_ERROR(Service_Audio, "Attempting to write past the end of output buffer!");
+            }
+            for (u32 i = 0; i < samples_read; i++) {
+                output_buffer[i] = temp_buffer[i];
+            }
+        } else {
+            std::memset(&temp_buffer[temp_buffer_pos], 0,
+                        (samples_to_read - samples_read) * sizeof(s16));
+
+            Resample(output_buffer, temp_buffer, sample_rate_ratio, fraction, samples_to_write,
+                     args.src_quality);
+
+            std::memcpy(voice_state.sample_history.data(), &temp_buffer[samples_to_read],
+                        pitch * sizeof(s16));
+        }
+
+        remaining_sample_count -= samples_to_write;
+        if (remaining_sample_count != 0 && is_buffer_starved) {
+            LOG_ERROR(Service_Audio, "Samples remaining but buffer is starving??");
+            break;
+        }
+
+        output_buffer = output_buffer.subspan(samples_to_write);
+    }
+
+    voice_state.wave_buffers_consumed = wavebuffers_consumed;
+    voice_state.played_sample_count = played_sample_count;
+    voice_state.wave_buffer_index = wavebuffer_index;
+    voice_state.offset = offset;
+    voice_state.fraction = fraction;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/data_source/decode.h

@@ -0,0 +1,59 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <span>
+
+#include "audio_core/common/common.h"
+#include "audio_core/common/wave_buffer.h"
+#include "audio_core/renderer/voice/voice_state.h"
+#include "common/common_types.h"
+
+namespace Core::Memory {
+class Memory;
+}
+
+namespace AudioCore::AudioRenderer {
+
+struct DecodeFromWaveBuffersArgs {
+    SampleFormat sample_format;
+    std::span<s32> output;
+    VoiceState* voice_state;
+    std::span<WaveBufferVersion2> wave_buffers;
+    s8 channel;
+    s8 channel_count;
+    SrcQuality src_quality;
+    f32 pitch;
+    u32 source_sample_rate;
+    u32 target_sample_rate;
+    u32 sample_count;
+    CpuAddr data_address;
+    u64 data_size;
+    bool IsVoicePlayedSampleCountResetAtLoopPointSupported;
+    bool IsVoicePitchAndSrcSkippedSupported;
+};
+
+struct DecodeArg {
+    CpuAddr buffer;
+    u64 buffer_size;
+    u32 start_offset;
+    u32 end_offset;
+    s8 channel_count;
+    std::array<s16, 16> coefficients;
+    VoiceState::AdpcmContext* adpcm_context;
+    s8 target_channel;
+    u32 offset;
+    u32 samples_to_read;
+};
+
+/**
+ * Decode wavebuffers according to the given args.
+ *
+ * @param memory - Core memory to read data from.
+ * @param args - The wavebuffer data, and information for how to decode it.
+ */
+void DecodeFromWaveBuffers(Core::Memory::Memory& memory, const DecodeFromWaveBuffersArgs& args);
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/data_source/pcm_float.cpp

@@ -0,0 +1,86 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/data_source/decode.h"
+#include "audio_core/renderer/command/data_source/pcm_float.h"
+
+namespace AudioCore::AudioRenderer {
+
+void PcmFloatDataSourceVersion1Command::Dump(const ADSP::CommandListProcessor& processor,
+                                             std::string& string) {
+    string +=
+        fmt::format("PcmFloatDataSourceVersion1Command\n\toutput_index {:02X} channel {} "
+                    "channel count {} source sample rate {} target sample rate {} src quality {}\n",
+                    output_index, channel_index, channel_count, sample_rate,
+                    processor.target_sample_rate, src_quality);
+}
+
+void PcmFloatDataSourceVersion1Command::Process(const ADSP::CommandListProcessor& processor) {
+    auto out_buffer = processor.mix_buffers.subspan(output_index * processor.sample_count,
+                                                    processor.sample_count);
+
+    DecodeFromWaveBuffersArgs args{
+        .sample_format{SampleFormat::PcmFloat},
+        .output{out_buffer},
+        .voice_state{reinterpret_cast<VoiceState*>(voice_state)},
+        .wave_buffers{wave_buffers},
+        .channel{channel_index},
+        .channel_count{channel_count},
+        .src_quality{src_quality},
+        .pitch{pitch},
+        .source_sample_rate{sample_rate},
+        .target_sample_rate{processor.target_sample_rate},
+        .sample_count{processor.sample_count},
+        .data_address{0},
+        .data_size{0},
+        .IsVoicePlayedSampleCountResetAtLoopPointSupported{(flags & 1) != 0},
+        .IsVoicePitchAndSrcSkippedSupported{(flags & 2) != 0},
+    };
+
+    DecodeFromWaveBuffers(*processor.memory, args);
+}
+
+bool PcmFloatDataSourceVersion1Command::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+void PcmFloatDataSourceVersion2Command::Dump(const ADSP::CommandListProcessor& processor,
+                                             std::string& string) {
+    string +=
+        fmt::format("PcmFloatDataSourceVersion2Command\n\toutput_index {:02X} channel {} "
+                    "channel count {} source sample rate {} target sample rate {} src quality {}\n",
+                    output_index, channel_index, channel_count, sample_rate,
+                    processor.target_sample_rate, src_quality);
+}
+
+void PcmFloatDataSourceVersion2Command::Process(const ADSP::CommandListProcessor& processor) {
+    auto out_buffer = processor.mix_buffers.subspan(output_index * processor.sample_count,
+                                                    processor.sample_count);
+
+    DecodeFromWaveBuffersArgs args{
+        .sample_format{SampleFormat::PcmFloat},
+        .output{out_buffer},
+        .voice_state{reinterpret_cast<VoiceState*>(voice_state)},
+        .wave_buffers{wave_buffers},
+        .channel{channel_index},
+        .channel_count{channel_count},
+        .src_quality{src_quality},
+        .pitch{pitch},
+        .source_sample_rate{sample_rate},
+        .target_sample_rate{processor.target_sample_rate},
+        .sample_count{processor.sample_count},
+        .data_address{0},
+        .data_size{0},
+        .IsVoicePlayedSampleCountResetAtLoopPointSupported{(flags & 1) != 0},
+        .IsVoicePitchAndSrcSkippedSupported{(flags & 2) != 0},
+    };
+
+    DecodeFromWaveBuffers(*processor.memory, args);
+}
+
+bool PcmFloatDataSourceVersion2Command::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/data_source/pcm_float.h

@@ -0,0 +1,113 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <string>
+
+#include "audio_core/common/wave_buffer.h"
+#include "audio_core/renderer/command/icommand.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command to decode PCM float-encoded version 1 wavebuffers
+ * into the output_index mix buffer.
+ */
+struct PcmFloatDataSourceVersion1Command : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Quality used for sample rate conversion
+    SrcQuality src_quality;
+    /// Mix buffer index for decoded samples
+    s16 output_index;
+    /// Flags to control decoding (see AudioCore::AudioRenderer::VoiceInfo::Flags)
+    u16 flags;
+    /// Wavebuffer sample rate
+    u32 sample_rate;
+    /// Pitch used for sample rate conversion
+    f32 pitch;
+    /// Target channel to read within the wavebuffer
+    s8 channel_index;
+    /// Number of channels within the wavebuffer
+    s8 channel_count;
+    /// Wavebuffers containing the wavebuffer address, context address, looping information etc
+    std::array<WaveBufferVersion2, MaxWaveBuffers> wave_buffers;
+    /// Voice state, updated each call and written back to game
+    CpuAddr voice_state;
+};
+
+/**
+ * AudioRenderer command to decode PCM float-encoded version 2 wavebuffers
+ * into the output_index mix buffer.
+ */
+struct PcmFloatDataSourceVersion2Command : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Quality used for sample rate conversion
+    SrcQuality src_quality;
+    /// Mix buffer index for decoded samples
+    s16 output_index;
+    /// Flags to control decoding (see AudioCore::AudioRenderer::VoiceInfo::Flags)
+    u16 flags;
+    /// Wavebuffer sample rate
+    u32 sample_rate;
+    /// Pitch used for sample rate conversion
+    f32 pitch;
+    /// Target channel to read within the wavebuffer
+    s8 channel_index;
+    /// Number of channels within the wavebuffer
+    s8 channel_count;
+    /// Wavebuffers containing the wavebuffer address, context address, looping information etc
+    std::array<WaveBufferVersion2, MaxWaveBuffers> wave_buffers;
+    /// Voice state, updated each call and written back to game
+    CpuAddr voice_state;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/data_source/pcm_int16.cpp

@@ -0,0 +1,87 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <span>
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/data_source/decode.h"
+#include "audio_core/renderer/command/data_source/pcm_int16.h"
+
+namespace AudioCore::AudioRenderer {
+
+void PcmInt16DataSourceVersion1Command::Dump(const ADSP::CommandListProcessor& processor,
+                                             std::string& string) {
+    string +=
+        fmt::format("PcmInt16DataSourceVersion1Command\n\toutput_index {:02X} channel {} "
+                    "channel count {} source sample rate {} target sample rate {} src quality {}\n",
+                    output_index, channel_index, channel_count, sample_rate,
+                    processor.target_sample_rate, src_quality);
+}
+
+void PcmInt16DataSourceVersion1Command::Process(const ADSP::CommandListProcessor& processor) {
+    auto out_buffer = processor.mix_buffers.subspan(output_index * processor.sample_count,
+                                                    processor.sample_count);
+
+    DecodeFromWaveBuffersArgs args{
+        .sample_format{SampleFormat::PcmInt16},
+        .output{out_buffer},
+        .voice_state{reinterpret_cast<VoiceState*>(voice_state)},
+        .wave_buffers{wave_buffers},
+        .channel{channel_index},
+        .channel_count{channel_count},
+        .src_quality{src_quality},
+        .pitch{pitch},
+        .source_sample_rate{sample_rate},
+        .target_sample_rate{processor.target_sample_rate},
+        .sample_count{processor.sample_count},
+        .data_address{0},
+        .data_size{0},
+        .IsVoicePlayedSampleCountResetAtLoopPointSupported{(flags & 1) != 0},
+        .IsVoicePitchAndSrcSkippedSupported{(flags & 2) != 0},
+    };
+
+    DecodeFromWaveBuffers(*processor.memory, args);
+}
+
+bool PcmInt16DataSourceVersion1Command::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+void PcmInt16DataSourceVersion2Command::Dump(const ADSP::CommandListProcessor& processor,
+                                             std::string& string) {
+    string +=
+        fmt::format("PcmInt16DataSourceVersion2Command\n\toutput_index {:02X} channel {} "
+                    "channel count {} source sample rate {} target sample rate {} src quality {}\n",
+                    output_index, channel_index, channel_count, sample_rate,
+                    processor.target_sample_rate, src_quality);
+}
+
+void PcmInt16DataSourceVersion2Command::Process(const ADSP::CommandListProcessor& processor) {
+    auto out_buffer = processor.mix_buffers.subspan(output_index * processor.sample_count,
+                                                    processor.sample_count);
+    DecodeFromWaveBuffersArgs args{
+        .sample_format{SampleFormat::PcmInt16},
+        .output{out_buffer},
+        .voice_state{reinterpret_cast<VoiceState*>(voice_state)},
+        .wave_buffers{wave_buffers},
+        .channel{channel_index},
+        .channel_count{channel_count},
+        .src_quality{src_quality},
+        .pitch{pitch},
+        .source_sample_rate{sample_rate},
+        .target_sample_rate{processor.target_sample_rate},
+        .sample_count{processor.sample_count},
+        .data_address{0},
+        .data_size{0},
+        .IsVoicePlayedSampleCountResetAtLoopPointSupported{(flags & 1) != 0},
+        .IsVoicePitchAndSrcSkippedSupported{(flags & 2) != 0},
+    };
+
+    DecodeFromWaveBuffers(*processor.memory, args);
+}
+
+bool PcmInt16DataSourceVersion2Command::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/data_source/pcm_int16.h

@@ -0,0 +1,110 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <string>
+
+#include "audio_core/common/wave_buffer.h"
+#include "audio_core/renderer/command/icommand.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command to decode PCM s16-encoded version 1 wavebuffers
+ * into the output_index mix buffer.
+ */
+struct PcmInt16DataSourceVersion1Command : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Quality used for sample rate conversion
+    SrcQuality src_quality;
+    /// Mix buffer index for decoded samples
+    s16 output_index;
+    /// Flags to control decoding (see AudioCore::AudioRenderer::VoiceInfo::Flags)
+    u16 flags;
+    /// Wavebuffer sample rate
+    u32 sample_rate;
+    /// Pitch used for sample rate conversion
+    f32 pitch;
+    /// Target channel to read within the wavebuffer
+    s8 channel_index;
+    /// Number of channels within the wavebuffer
+    s8 channel_count;
+    /// Wavebuffers containing the wavebuffer address, context address, looping information etc
+    std::array<WaveBufferVersion2, MaxWaveBuffers> wave_buffers;
+    /// Voice state, updated each call and written back to game
+    CpuAddr voice_state;
+};
+
+/**
+ * AudioRenderer command to decode PCM s16-encoded version 2 wavebuffers
+ * into the output_index mix buffer.
+ */
+struct PcmInt16DataSourceVersion2Command : ICommand {
+    /**
+     * Print this command's information to a string.
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Quality used for sample rate conversion
+    SrcQuality src_quality;
+    /// Mix buffer index for decoded samples
+    s16 output_index;
+    /// Flags to control decoding (see AudioCore::AudioRenderer::VoiceInfo::Flags)
+    u16 flags;
+    /// Wavebuffer sample rate
+    u32 sample_rate;
+    /// Pitch used for sample rate conversion
+    f32 pitch;
+    /// Target channel to read within the wavebuffer
+    s8 channel_index;
+    /// Number of channels within the wavebuffer
+    s8 channel_count;
+    /// Wavebuffers containing the wavebuffer address, context address, looping information etc
+    std::array<WaveBufferVersion2, MaxWaveBuffers> wave_buffers;
+    /// Voice state, updated each call and written back to game
+    CpuAddr voice_state;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/aux_.cpp

@@ -0,0 +1,207 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/effect/aux_.h"
+#include "audio_core/renderer/effect/effect_aux_info.h"
+#include "core/memory.h"
+
+namespace AudioCore::AudioRenderer {
+/**
+ * Reset an AuxBuffer.
+ *
+ * @param memory   - Core memory for writing.
+ * @param aux_info - Memory address pointing to the AuxInfo to reset.
+ */
+static void ResetAuxBufferDsp(Core::Memory::Memory& memory, const CpuAddr aux_info) {
+    if (aux_info == 0) {
+        LOG_ERROR(Service_Audio, "Aux info is 0!");
+        return;
+    }
+
+    auto info{reinterpret_cast<AuxInfo::AuxInfoDsp*>(memory.GetPointer(aux_info))};
+    info->read_offset = 0;
+    info->write_offset = 0;
+    info->total_sample_count = 0;
+}
+
+/**
+ * Write the given input mix buffer to the memory at send_buffer, and update send_info_ if
+ * update_count is set, to notify the game that an update happened.
+ *
+ * @param memory       - Core memory for writing.
+ * @param send_info_   - Meta information for where to write the mix buffer.
+ * @param sample_count - Unused.
+ * @param send_buffer  - Memory address to write the mix buffer to.
+ * @param count_max    - Maximum number of samples in the receiving buffer.
+ * @param input        - Input mix buffer to write.
+ * @param write_count_ - Number of samples to write.
+ * @param write_offset - Current offset to begin writing the receiving buffer at.
+ * @param update_count - If non-zero, send_info_ will be updated.
+ * @return Number of samples written.
+ */
+static u32 WriteAuxBufferDsp(Core::Memory::Memory& memory, const CpuAddr send_info_,
+                             [[maybe_unused]] u32 sample_count, const CpuAddr send_buffer,
+                             const u32 count_max, std::span<const s32> input,
+                             const u32 write_count_, const u32 write_offset,
+                             const u32 update_count) {
+    if (write_count_ > count_max) {
+        LOG_ERROR(Service_Audio,
+                  "write_count must be smaller than count_max! write_count {}, count_max {}",
+                  write_count_, count_max);
+        return 0;
+    }
+
+    if (input.empty()) {
+        LOG_ERROR(Service_Audio, "input buffer is empty!");
+        return 0;
+    }
+
+    if (send_buffer == 0) {
+        LOG_ERROR(Service_Audio, "send_buffer is 0!");
+        return 0;
+    }
+
+    if (count_max == 0) {
+        return 0;
+    }
+
+    AuxInfo::AuxInfoDsp send_info{};
+    memory.ReadBlockUnsafe(send_info_, &send_info, sizeof(AuxInfo::AuxInfoDsp));
+
+    u32 target_write_offset{send_info.write_offset + write_offset};
+    if (target_write_offset > count_max || write_count_ == 0) {
+        return 0;
+    }
+
+    u32 write_count{write_count_};
+    u32 write_pos{0};
+    while (write_count > 0) {
+        u32 to_write{std::min(count_max - target_write_offset, write_count)};
+
+        if (to_write > 0) {
+            memory.WriteBlockUnsafe(send_buffer + target_write_offset * sizeof(s32),
+                                    &input[write_pos], to_write * sizeof(s32));
+        }
+
+        target_write_offset = (target_write_offset + to_write) % count_max;
+        write_count -= to_write;
+        write_pos += to_write;
+    }
+
+    if (update_count) {
+        send_info.write_offset = (send_info.write_offset + update_count) % count_max;
+    }
+
+    memory.WriteBlockUnsafe(send_info_, &send_info, sizeof(AuxInfo::AuxInfoDsp));
+
+    return write_count_;
+}
+
+/**
+ * Read the given memory at return_buffer into the output mix buffer, and update return_info_ if
+ * update_count is set, to notify the game that an update happened.
+ *
+ * @param memory        - Core memory for writing.
+ * @param return_info_  - Meta information for where to read the mix buffer.
+ * @param return_buffer - Memory address to read the samples from.
+ * @param count_max     - Maximum number of samples in the receiving buffer.
+ * @param output        - Output mix buffer which will receive the samples.
+ * @param count_        - Number of samples to read.
+ * @param read_offset   - Current offset to begin reading the return_buffer at.
+ * @param update_count  - If non-zero, send_info_ will be updated.
+ * @return Number of samples read.
+ */
+static u32 ReadAuxBufferDsp(Core::Memory::Memory& memory, const CpuAddr return_info_,
+                            const CpuAddr return_buffer, const u32 count_max, std::span<s32> output,
+                            const u32 count_, const u32 read_offset, const u32 update_count) {
+    if (count_max == 0) {
+        return 0;
+    }
+
+    if (count_ > count_max) {
+        LOG_ERROR(Service_Audio, "count must be smaller than count_max! count {}, count_max {}",
+                  count_, count_max);
+        return 0;
+    }
+
+    if (output.empty()) {
+        LOG_ERROR(Service_Audio, "output buffer is empty!");
+        return 0;
+    }
+
+    if (return_buffer == 0) {
+        LOG_ERROR(Service_Audio, "return_buffer is 0!");
+        return 0;
+    }
+
+    AuxInfo::AuxInfoDsp return_info{};
+    memory.ReadBlockUnsafe(return_info_, &return_info, sizeof(AuxInfo::AuxInfoDsp));
+
+    u32 target_read_offset{return_info.read_offset + read_offset};
+    if (target_read_offset > count_max) {
+        return 0;
+    }
+
+    u32 read_count{count_};
+    u32 read_pos{0};
+    while (read_count > 0) {
+        u32 to_read{std::min(count_max - target_read_offset, read_count)};
+
+        if (to_read > 0) {
+            memory.ReadBlockUnsafe(return_buffer + target_read_offset * sizeof(s32),
+                                   &output[read_pos], to_read * sizeof(s32));
+        }
+
+        target_read_offset = (target_read_offset + to_read) % count_max;
+        read_count -= to_read;
+        read_pos += to_read;
+    }
+
+    if (update_count) {
+        return_info.read_offset = (return_info.read_offset + update_count) % count_max;
+    }
+
+    memory.WriteBlockUnsafe(return_info_, &return_info, sizeof(AuxInfo::AuxInfoDsp));
+
+    return count_;
+}
+
+void AuxCommand::Dump([[maybe_unused]] const ADSP::CommandListProcessor& processor,
+                      std::string& string) {
+    string += fmt::format("AuxCommand\n\tenabled {} input {:02X} output {:02X}\n", effect_enabled,
+                          input, output);
+}
+
+void AuxCommand::Process(const ADSP::CommandListProcessor& processor) {
+    auto input_buffer{
+        processor.mix_buffers.subspan(input * processor.sample_count, processor.sample_count)};
+    auto output_buffer{
+        processor.mix_buffers.subspan(output * processor.sample_count, processor.sample_count)};
+
+    if (effect_enabled) {
+        WriteAuxBufferDsp(*processor.memory, send_buffer_info, processor.sample_count, send_buffer,
+                          count_max, input_buffer, processor.sample_count, write_offset,
+                          update_count);
+
+        auto read{ReadAuxBufferDsp(*processor.memory, return_buffer_info, return_buffer, count_max,
+                                   output_buffer, processor.sample_count, write_offset,
+                                   update_count)};
+
+        if (read != processor.sample_count) {
+            std::memset(&output_buffer[read], 0, processor.sample_count - read);
+        }
+    } else {
+        ResetAuxBufferDsp(*processor.memory, send_buffer_info);
+        ResetAuxBufferDsp(*processor.memory, return_buffer_info);
+        if (input != output) {
+            std::memcpy(output_buffer.data(), input_buffer.data(), output_buffer.size_bytes());
+        }
+    }
+}
+
+bool AuxCommand::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/aux_.h

@@ -0,0 +1,66 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <string>
+
+#include "audio_core/renderer/command/icommand.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command to read and write an auxiliary buffer, writing the input mix buffer to game
+ * memory, and reading into the output buffer from game memory.
+ */
+struct AuxCommand : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Input mix buffer index
+    s16 input;
+    /// Output mix buffer index
+    s16 output;
+    /// Meta info for writing
+    CpuAddr send_buffer_info;
+    /// Meta info for reading
+    CpuAddr return_buffer_info;
+    /// Game memory write buffer
+    CpuAddr send_buffer;
+    /// Game memory read buffer
+    CpuAddr return_buffer;
+    /// Max samples to read/write
+    u32 count_max;
+    /// Current read/write offset
+    u32 write_offset;
+    /// Number of samples to update per call
+    u32 update_count;
+    /// is this effect enabled?
+    bool effect_enabled;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/biquad_filter.cpp

@@ -0,0 +1,118 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/effect/biquad_filter.h"
+#include "audio_core/renderer/voice/voice_state.h"
+
+namespace AudioCore::AudioRenderer {
+/**
+ * Biquad filter float implementation.
+ *
+ * @param output       - Output container for filtered samples.
+ * @param input        - Input container for samples to be filtered.
+ * @param b            - Feedforward coefficients.
+ * @param a            - Feedback coefficients.
+ * @param state        - State to track previous samples between calls.
+ * @param sample_count - Number of samples to process.
+ */
+void ApplyBiquadFilterFloat(std::span<s32> output, std::span<const s32> input,
+                            std::array<s16, 3>& b_, std::array<s16, 2>& a_,
+                            VoiceState::BiquadFilterState& state, const u32 sample_count) {
+    constexpr s64 min{std::numeric_limits<s32>::min()};
+    constexpr s64 max{std::numeric_limits<s32>::max()};
+    std::array<f64, 3> b{Common::FixedPoint<50, 14>::from_base(b_[0]).to_double(),
+                         Common::FixedPoint<50, 14>::from_base(b_[1]).to_double(),
+                         Common::FixedPoint<50, 14>::from_base(b_[2]).to_double()};
+    std::array<f64, 2> a{Common::FixedPoint<50, 14>::from_base(a_[0]).to_double(),
+                         Common::FixedPoint<50, 14>::from_base(a_[1]).to_double()};
+    std::array<f64, 4> s{state.s0.to_double(), state.s1.to_double(), state.s2.to_double(),
+                         state.s3.to_double()};
+
+    for (u32 i = 0; i < sample_count; i++) {
+        f64 in_sample{static_cast<f64>(input[i])};
+        auto sample{in_sample * b[0] + s[0] * b[1] + s[1] * b[2] + s[2] * a[0] + s[3] * a[1]};
+
+        output[i] = static_cast<s32>(std::clamp(static_cast<s64>(sample), min, max));
+
+        s[1] = s[0];
+        s[0] = in_sample;
+        s[3] = s[2];
+        s[2] = sample;
+    }
+
+    state.s0 = s[0];
+    state.s1 = s[1];
+    state.s2 = s[2];
+    state.s3 = s[3];
+}
+
+/**
+ * Biquad filter s32 implementation.
+ *
+ * @param output       - Output container for filtered samples.
+ * @param input        - Input container for samples to be filtered.
+ * @param b            - Feedforward coefficients.
+ * @param a            - Feedback coefficients.
+ * @param state        - State to track previous samples between calls.
+ * @param sample_count - Number of samples to process.
+ */
+static void ApplyBiquadFilterInt(std::span<s32> output, std::span<const s32> input,
+                                 std::array<s16, 3>& b_, std::array<s16, 2>& a_,
+                                 VoiceState::BiquadFilterState& state, const u32 sample_count) {
+    constexpr s64 min{std::numeric_limits<s32>::min()};
+    constexpr s64 max{std::numeric_limits<s32>::max()};
+    std::array<Common::FixedPoint<50, 14>, 3> b{
+        Common::FixedPoint<50, 14>::from_base(b_[0]),
+        Common::FixedPoint<50, 14>::from_base(b_[1]),
+        Common::FixedPoint<50, 14>::from_base(b_[2]),
+    };
+    std::array<Common::FixedPoint<50, 14>, 3> a{
+        Common::FixedPoint<50, 14>::from_base(a_[0]),
+        Common::FixedPoint<50, 14>::from_base(a_[1]),
+    };
+
+    for (u32 i = 0; i < sample_count; i++) {
+        s64 in_sample{input[i]};
+        auto sample{in_sample * b[0] + state.s0};
+        const auto out_sample{std::clamp(sample.to_long(), min, max)};
+
+        output[i] = static_cast<s32>(out_sample);
+
+        state.s0 = state.s1 + b[1] * in_sample + a[0] * out_sample;
+        state.s1 = 0 + b[2] * in_sample + a[1] * out_sample;
+    }
+}
+
+void BiquadFilterCommand::Dump([[maybe_unused]] const ADSP::CommandListProcessor& processor,
+                               std::string& string) {
+    string += fmt::format(
+        "BiquadFilterCommand\n\tinput {:02X} output {:02X} needs_init {} use_float_processing {}\n",
+        input, output, needs_init, use_float_processing);
+}
+
+void BiquadFilterCommand::Process(const ADSP::CommandListProcessor& processor) {
+    auto state_{reinterpret_cast<VoiceState::BiquadFilterState*>(state)};
+    if (needs_init) {
+        std::memset(state_, 0, sizeof(VoiceState::BiquadFilterState));
+    }
+
+    auto input_buffer{
+        processor.mix_buffers.subspan(input * processor.sample_count, processor.sample_count)};
+    auto output_buffer{
+        processor.mix_buffers.subspan(output * processor.sample_count, processor.sample_count)};
+
+    if (use_float_processing) {
+        ApplyBiquadFilterFloat(output_buffer, input_buffer, biquad.b, biquad.a, *state_,
+                               processor.sample_count);
+    } else {
+        ApplyBiquadFilterInt(output_buffer, input_buffer, biquad.b, biquad.a, *state_,
+                             processor.sample_count);
+    }
+}
+
+bool BiquadFilterCommand::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/biquad_filter.h

@@ -0,0 +1,74 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <string>
+
+#include "audio_core/renderer/command/icommand.h"
+#include "audio_core/renderer/voice/voice_info.h"
+#include "audio_core/renderer/voice/voice_state.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command for applying a biquad filter to the input mix buffer, saving the results to
+ * the output mix buffer.
+ */
+struct BiquadFilterCommand : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Input mix buffer index
+    s16 input;
+    /// Output mix buffer index
+    s16 output;
+    /// Input parameters for biquad
+    VoiceInfo::BiquadFilterParameter biquad;
+    /// Biquad state, updated each call
+    CpuAddr state;
+    /// If true, reset the state
+    bool needs_init;
+    /// If true, use float processing rather than int
+    bool use_float_processing;
+};
+
+/**
+ * Biquad filter float implementation.
+ *
+ * @param output       - Output container for filtered samples.
+ * @param input        - Input container for samples to be filtered.
+ * @param b            - Feedforward coefficients.
+ * @param a            - Feedback coefficients.
+ * @param state        - State to track previous samples.
+ * @param sample_count - Number of samples to process.
+ */
+void ApplyBiquadFilterFloat(std::span<s32> output, std::span<const s32> input,
+                            std::array<s16, 3>& b, std::array<s16, 2>& a,
+                            VoiceState::BiquadFilterState& state, const u32 sample_count);
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/capture.cpp

@@ -0,0 +1,142 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/effect/capture.h"
+#include "audio_core/renderer/effect/effect_aux_info.h"
+#include "core/memory.h"
+
+namespace AudioCore::AudioRenderer {
+/**
+ * Reset an AuxBuffer.
+ *
+ * @param memory   - Core memory for writing.
+ * @param aux_info - Memory address pointing to the AuxInfo to reset.
+ */
+static void ResetAuxBufferDsp(Core::Memory::Memory& memory, const CpuAddr aux_info) {
+    if (aux_info == 0) {
+        LOG_ERROR(Service_Audio, "Aux info is 0!");
+        return;
+    }
+
+    memory.Write32(VAddr(aux_info + offsetof(AuxInfo::AuxInfoDsp, read_offset)), 0);
+    memory.Write32(VAddr(aux_info + offsetof(AuxInfo::AuxInfoDsp, write_offset)), 0);
+    memory.Write32(VAddr(aux_info + offsetof(AuxInfo::AuxInfoDsp, total_sample_count)), 0);
+}
+
+/**
+ * Write the given input mix buffer to the memory at send_buffer, and update send_info_ if
+ * update_count is set, to notify the game that an update happened.
+ *
+ * @param memory       - Core memory for writing.
+ * @param send_info_   - Header information for where to write the mix buffer.
+ * @param send_buffer  - Memory address to write the mix buffer to.
+ * @param count_max    - Maximum number of samples in the receiving buffer.
+ * @param input        - Input mix buffer to write.
+ * @param write_count_ - Number of samples to write.
+ * @param write_offset - Current offset to begin writing the receiving buffer at.
+ * @param update_count - If non-zero, send_info_ will be updated.
+ * @return Number of samples written.
+ */
+static u32 WriteAuxBufferDsp(Core::Memory::Memory& memory, const CpuAddr send_info_,
+                             const CpuAddr send_buffer, u32 count_max, std::span<const s32> input,
+                             const u32 write_count_, const u32 write_offset,
+                             const u32 update_count) {
+    if (write_count_ > count_max) {
+        LOG_ERROR(Service_Audio,
+                  "write_count must be smaller than count_max! write_count {}, count_max {}",
+                  write_count_, count_max);
+        return 0;
+    }
+
+    if (send_info_ == 0) {
+        LOG_ERROR(Service_Audio, "send_info is 0!");
+        return 0;
+    }
+
+    if (input.empty()) {
+        LOG_ERROR(Service_Audio, "input buffer is empty!");
+        return 0;
+    }
+
+    if (send_buffer == 0) {
+        LOG_ERROR(Service_Audio, "send_buffer is 0!");
+        return 0;
+    }
+
+    if (count_max == 0) {
+        return 0;
+    }
+
+    AuxInfo::AuxBufferInfo send_info{};
+    memory.ReadBlockUnsafe(send_info_, &send_info, sizeof(AuxInfo::AuxBufferInfo));
+
+    u32 target_write_offset{send_info.dsp_info.write_offset + write_offset};
+    if (target_write_offset > count_max || write_count_ == 0) {
+        return 0;
+    }
+
+    u32 write_count{write_count_};
+    u32 write_pos{0};
+    while (write_count > 0) {
+        u32 to_write{std::min(count_max - target_write_offset, write_count)};
+
+        if (to_write > 0) {
+            memory.WriteBlockUnsafe(send_buffer + target_write_offset * sizeof(s32),
+                                    &input[write_pos], to_write * sizeof(s32));
+        }
+
+        target_write_offset = (target_write_offset + to_write) % count_max;
+        write_count -= to_write;
+        write_pos += to_write;
+    }
+
+    if (update_count) {
+        const auto count_diff{send_info.dsp_info.total_sample_count -
+                              send_info.cpu_info.total_sample_count};
+        if (count_diff >= count_max) {
+            auto dsp_lost_count{send_info.dsp_info.lost_sample_count + update_count};
+            if (dsp_lost_count - send_info.cpu_info.lost_sample_count <
+                send_info.dsp_info.lost_sample_count - send_info.cpu_info.lost_sample_count) {
+                dsp_lost_count = send_info.cpu_info.lost_sample_count - 1;
+            }
+            send_info.dsp_info.lost_sample_count = dsp_lost_count;
+        }
+
+        send_info.dsp_info.write_offset =
+            (send_info.dsp_info.write_offset + update_count + count_max) % count_max;
+
+        auto new_sample_count{send_info.dsp_info.total_sample_count + update_count};
+        if (new_sample_count - send_info.cpu_info.total_sample_count < count_diff) {
+            new_sample_count = send_info.cpu_info.total_sample_count - 1;
+        }
+        send_info.dsp_info.total_sample_count = new_sample_count;
+    }
+
+    memory.WriteBlockUnsafe(send_info_, &send_info, sizeof(AuxInfo::AuxBufferInfo));
+
+    return write_count_;
+}
+
+void CaptureCommand::Dump([[maybe_unused]] const ADSP::CommandListProcessor& processor,
+                          std::string& string) {
+    string += fmt::format("CaptureCommand\n\tenabled {} input {:02X} output {:02X}", effect_enabled,
+                          input, output);
+}
+
+void CaptureCommand::Process(const ADSP::CommandListProcessor& processor) {
+    if (effect_enabled) {
+        auto input_buffer{
+            processor.mix_buffers.subspan(input * processor.sample_count, processor.sample_count)};
+        WriteAuxBufferDsp(*processor.memory, send_buffer_info, send_buffer, count_max, input_buffer,
+                          processor.sample_count, write_offset, update_count);
+    } else {
+        ResetAuxBufferDsp(*processor.memory, send_buffer_info);
+    }
+}
+
+bool CaptureCommand::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/capture.h

@@ -0,0 +1,62 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <string>
+
+#include "audio_core/renderer/command/icommand.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command for capturing a mix buffer. That is, writing it back to a given game memory
+ * address.
+ */
+struct CaptureCommand : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Input mix buffer index
+    s16 input;
+    /// Output mix buffer index
+    s16 output;
+    /// Meta info for writing
+    CpuAddr send_buffer_info;
+    /// Game memory write buffer
+    CpuAddr send_buffer;
+    /// Max samples to read/write
+    u32 count_max;
+    /// Current read/write offset
+    u32 write_offset;
+    /// Number of samples to update per call
+    u32 update_count;
+    /// is this effect enabled?
+    bool effect_enabled;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/delay.cpp

@@ -0,0 +1,244 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/effect/delay.h"
+
+namespace AudioCore::AudioRenderer {
+/**
+ * Update the DelayInfo state according to the given parameters.
+ *
+ * @param params - Input parameters to update the state.
+ * @param state  - State to be updated.
+ */
+static void SetDelayEffectParameter(const DelayInfo::ParameterVersion1& params,
+                                    DelayInfo::State& state) {
+    auto channel_spread{params.channel_spread};
+    state.feedback_gain = params.feedback_gain * 0.97998046875f;
+    state.delay_feedback_gain = state.feedback_gain * (1.0f - channel_spread);
+    if (params.channel_count == 4 || params.channel_count == 6) {
+        channel_spread >>= 1;
+    }
+    state.delay_feedback_cross_gain = channel_spread * state.feedback_gain;
+    state.lowpass_feedback_gain = params.lowpass_amount * 0.949951171875f;
+    state.lowpass_gain = 1.0f - state.lowpass_feedback_gain;
+}
+
+/**
+ * Initialize a new DelayInfo state according to the given parameters.
+ *
+ * @param params     - Input parameters to update the state.
+ * @param state      - State to be updated.
+ * @param workbuffer - Game-supplied memory for the state. (Unused)
+ */
+static void InitializeDelayEffect(const DelayInfo::ParameterVersion1& params,
+                                  DelayInfo::State& state,
+                                  [[maybe_unused]] const CpuAddr workbuffer) {
+    state = {};
+
+    for (u32 channel = 0; channel < params.channel_count; channel++) {
+        Common::FixedPoint<32, 32> sample_count_max{0.064f};
+        sample_count_max *= params.sample_rate.to_int_floor() * params.delay_time_max;
+
+        Common::FixedPoint<18, 14> delay_time{params.delay_time};
+        delay_time *= params.sample_rate / 1000;
+        Common::FixedPoint<32, 32> sample_count{delay_time};
+
+        if (sample_count > sample_count_max) {
+            sample_count = sample_count_max;
+        }
+
+        state.delay_lines[channel].sample_count_max = sample_count_max.to_int_floor();
+        state.delay_lines[channel].sample_count = sample_count.to_int_floor();
+        state.delay_lines[channel].buffer.resize(state.delay_lines[channel].sample_count, 0);
+        if (state.delay_lines[channel].buffer.size() == 0) {
+            state.delay_lines[channel].buffer.push_back(0);
+        }
+        state.delay_lines[channel].buffer_pos = 0;
+        state.delay_lines[channel].decay_rate = 1.0f;
+    }
+
+    SetDelayEffectParameter(params, state);
+}
+
+/**
+ * Delay effect impl, according to the parameters and current state, on the input mix buffers,
+ * saving the results to the output mix buffers.
+ *
+ * @param params       - Input parameters to use.
+ * @param state        - State to use, must be initialized (see InitializeDelayEffect).
+ * @param inputs       - Input mix buffers to performan the delay on.
+ * @param outputs      - Output mix buffers to receive the delayed samples.
+ * @param sample_count - Number of samples to process.
+ */
+template <size_t Channels>
+static void ApplyDelay(const DelayInfo::ParameterVersion1& params, DelayInfo::State& state,
+                       std::vector<std::span<const s32>>& inputs,
+                       std::vector<std::span<s32>>& outputs, const u32 sample_count) {
+    for (u32 i = 0; i < sample_count; i++) {
+        std::array<Common::FixedPoint<50, 14>, Channels> input_samples{};
+        for (u32 channel = 0; channel < Channels; channel++) {
+            input_samples[channel] = inputs[channel][i] * 64;
+        }
+
+        std::array<Common::FixedPoint<50, 14>, Channels> delay_samples{};
+        for (u32 channel = 0; channel < Channels; channel++) {
+            delay_samples[channel] = state.delay_lines[channel].Read();
+        }
+
+        std::array<std::array<Common::FixedPoint<18, 14>, Channels>, Channels> matrix{};
+        if constexpr (Channels == 1) {
+            matrix = {{
+                {state.feedback_gain},
+            }};
+        } else if constexpr (Channels == 2) {
+            matrix = {{
+                {state.delay_feedback_gain, state.delay_feedback_cross_gain},
+                {state.delay_feedback_cross_gain, state.delay_feedback_gain},
+            }};
+        } else if constexpr (Channels == 4) {
+            matrix = {{
+                {state.delay_feedback_gain, state.delay_feedback_cross_gain,
+                 state.delay_feedback_cross_gain, 0.0f},
+                {state.delay_feedback_cross_gain, state.delay_feedback_gain, 0.0f,
+                 state.delay_feedback_cross_gain},
+                {state.delay_feedback_cross_gain, 0.0f, state.delay_feedback_gain,
+                 state.delay_feedback_cross_gain},
+                {0.0f, state.delay_feedback_cross_gain, state.delay_feedback_cross_gain,
+                 state.delay_feedback_gain},
+            }};
+        } else if constexpr (Channels == 6) {
+            matrix = {{
+                {state.delay_feedback_gain, 0.0f, 0.0f, 0.0f, state.delay_feedback_cross_gain,
+                 state.delay_feedback_cross_gain},
+                {0.0f, state.delay_feedback_gain, 0.0f, state.delay_feedback_cross_gain,
+                 state.delay_feedback_cross_gain, 0.0f},
+                {state.delay_feedback_cross_gain, 0.0f, state.delay_feedback_gain,
+                 state.delay_feedback_cross_gain, 0.0f, 0.0f},
+                {0.0f, state.delay_feedback_cross_gain, state.delay_feedback_cross_gain,
+                 state.delay_feedback_gain, 0.0f, 0.0f},
+                {state.delay_feedback_cross_gain, state.delay_feedback_cross_gain, 0.0f, 0.0f,
+                 state.delay_feedback_gain, 0.0f},
+                {0.0f, 0.0f, 0.0f, 0.0f, 0.0f, params.feedback_gain},
+            }};
+        }
+
+        std::array<Common::FixedPoint<50, 14>, Channels> gained_samples{};
+        for (u32 channel = 0; channel < Channels; channel++) {
+            Common::FixedPoint<50, 14> delay{};
+            for (u32 j = 0; j < Channels; j++) {
+                delay += delay_samples[j] * matrix[j][channel];
+            }
+            gained_samples[channel] = input_samples[channel] * params.in_gain + delay;
+        }
+
+        for (u32 channel = 0; channel < Channels; channel++) {
+            state.lowpass_z[channel] = gained_samples[channel] * state.lowpass_gain +
+                                       state.lowpass_z[channel] * state.lowpass_feedback_gain;
+            state.delay_lines[channel].Write(state.lowpass_z[channel]);
+        }
+
+        for (u32 channel = 0; channel < Channels; channel++) {
+            outputs[channel][i] = (input_samples[channel] * params.dry_gain +
+                                   delay_samples[channel] * params.wet_gain)
+                                      .to_int_floor() /
+                                  64;
+        }
+    }
+}
+
+/**
+ * Apply a delay effect if enabled, according to the parameters and current state, on the input mix
+ * buffers, saving the results to the output mix buffers.
+ *
+ * @param params       - Input parameters to use.
+ * @param state        - State to use, must be initialized (see InitializeDelayEffect).
+ * @param enabled      - If enabled, delay will be applied, otherwise input is copied to output.
+ * @param inputs       - Input mix buffers to performan the delay on.
+ * @param outputs      - Output mix buffers to receive the delayed samples.
+ * @param sample_count - Number of samples to process.
+ */
+static void ApplyDelayEffect(const DelayInfo::ParameterVersion1& params, DelayInfo::State& state,
+                             const bool enabled, std::vector<std::span<const s32>>& inputs,
+                             std::vector<std::span<s32>>& outputs, const u32 sample_count) {
+
+    if (!IsChannelCountValid(params.channel_count)) {
+        LOG_ERROR(Service_Audio, "Invalid delay channels {}", params.channel_count);
+        return;
+    }
+
+    if (enabled) {
+        switch (params.channel_count) {
+        case 1:
+            ApplyDelay<1>(params, state, inputs, outputs, sample_count);
+            break;
+        case 2:
+            ApplyDelay<2>(params, state, inputs, outputs, sample_count);
+            break;
+        case 4:
+            ApplyDelay<4>(params, state, inputs, outputs, sample_count);
+            break;
+        case 6:
+            ApplyDelay<6>(params, state, inputs, outputs, sample_count);
+            break;
+        default:
+            for (u32 channel = 0; channel < params.channel_count; channel++) {
+                if (inputs[channel].data() != outputs[channel].data()) {
+                    std::memcpy(outputs[channel].data(), inputs[channel].data(),
+                                sample_count * sizeof(s32));
+                }
+            }
+            break;
+        }
+    } else {
+        for (u32 channel = 0; channel < params.channel_count; channel++) {
+            if (inputs[channel].data() != outputs[channel].data()) {
+                std::memcpy(outputs[channel].data(), inputs[channel].data(),
+                            sample_count * sizeof(s32));
+            }
+        }
+    }
+}
+
+void DelayCommand::Dump([[maybe_unused]] const ADSP::CommandListProcessor& processor,
+                        std::string& string) {
+    string += fmt::format("DelayCommand\n\tenabled {} \n\tinputs: ", effect_enabled);
+    for (u32 i = 0; i < MaxChannels; i++) {
+        string += fmt::format("{:02X}, ", inputs[i]);
+    }
+    string += "\n\toutputs: ";
+    for (u32 i = 0; i < MaxChannels; i++) {
+        string += fmt::format("{:02X}, ", outputs[i]);
+    }
+    string += "\n";
+}
+
+void DelayCommand::Process(const ADSP::CommandListProcessor& processor) {
+    std::vector<std::span<const s32>> input_buffers(parameter.channel_count);
+    std::vector<std::span<s32>> output_buffers(parameter.channel_count);
+
+    for (s16 i = 0; i < parameter.channel_count; i++) {
+        input_buffers[i] = processor.mix_buffers.subspan(inputs[i] * processor.sample_count,
+                                                         processor.sample_count);
+        output_buffers[i] = processor.mix_buffers.subspan(outputs[i] * processor.sample_count,
+                                                          processor.sample_count);
+    }
+
+    auto state_{reinterpret_cast<DelayInfo::State*>(state)};
+
+    if (effect_enabled) {
+        if (parameter.state == DelayInfo::ParameterState::Updating) {
+            SetDelayEffectParameter(parameter, *state_);
+        } else if (parameter.state == DelayInfo::ParameterState::Initialized) {
+            InitializeDelayEffect(parameter, *state_, workbuffer);
+        }
+    }
+    ApplyDelayEffect(parameter, *state_, effect_enabled, input_buffers, output_buffers,
+                     processor.sample_count);
+}
+
+bool DelayCommand::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/delay.h

@@ -0,0 +1,60 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <string>
+
+#include "audio_core/renderer/command/icommand.h"
+#include "audio_core/renderer/effect/effect_delay_info.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command for a delay effect. Delays inputs mix buffers according to the parameters
+ * and state, outputs receives the delayed samples.
+ */
+struct DelayCommand : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Input mix buffer offsets for each channel
+    std::array<s8, MaxChannels> inputs;
+    /// Output mix buffer offsets for each channel
+    std::array<s8, MaxChannels> outputs;
+    /// Input parameters
+    DelayInfo::ParameterVersion1 parameter;
+    /// State, updated each call
+    CpuAddr state;
+    /// Game-supplied workbuffer (Unused)
+    CpuAddr workbuffer;
+    /// Is this effect enabled?
+    bool effect_enabled;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/i3dl2_reverb.cpp

@@ -0,0 +1,428 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <numbers>
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/effect/i3dl2_reverb.h"
+
+namespace AudioCore::AudioRenderer {
+
+constexpr std::array<f32, I3dl2ReverbInfo::MaxDelayLines> MinDelayLineTimes{
+    5.0f,
+    6.0f,
+    13.0f,
+    14.0f,
+};
+constexpr std::array<f32, I3dl2ReverbInfo::MaxDelayLines> MaxDelayLineTimes{
+    45.7042007446f,
+    82.7817001343f,
+    149.938293457f,
+    271.575805664f,
+};
+constexpr std::array<f32, I3dl2ReverbInfo::MaxDelayLines> Decay0MaxDelayLineTimes{17.0f, 13.0f,
+                                                                                  9.0f, 7.0f};
+constexpr std::array<f32, I3dl2ReverbInfo::MaxDelayLines> Decay1MaxDelayLineTimes{19.0f, 11.0f,
+                                                                                  10.0f, 6.0f};
+constexpr std::array<f32, I3dl2ReverbInfo::MaxDelayTaps> EarlyTapTimes{
+    0.0171360000968f,
+    0.0591540001333f,
+    0.161733001471f,
+    0.390186011791f,
+    0.425262004137f,
+    0.455410987139f,
+    0.689737021923f,
+    0.74590998888f,
+    0.833844006062f,
+    0.859502017498f,
+    0.0f,
+    0.0750240013003f,
+    0.168788000941f,
+    0.299901008606f,
+    0.337442994118f,
+    0.371903002262f,
+    0.599011003971f,
+    0.716741025448f,
+    0.817858994007f,
+    0.85166400671f,
+};
+
+constexpr std::array<f32, I3dl2ReverbInfo::MaxDelayTaps> EarlyGains{
+    0.67096f, 0.61027f, 1.0f,     0.3568f,  0.68361f, 0.65978f, 0.51939f,
+    0.24712f, 0.45945f, 0.45021f, 0.64196f, 0.54879f, 0.92925f, 0.3827f,
+    0.72867f, 0.69794f, 0.5464f,  0.24563f, 0.45214f, 0.44042f};
+
+/**
+ * Update the I3dl2ReverbInfo state according to the given parameters.
+ *
+ * @param params - Input parameters to update the state.
+ * @param state  - State to be updated.
+ * @param reset  - If enabled, the state buffers will be reset. Only set this on initialize.
+ */
+static void UpdateI3dl2ReverbEffectParameter(const I3dl2ReverbInfo::ParameterVersion1& params,
+                                             I3dl2ReverbInfo::State& state, const bool reset) {
+    const auto pow_10 = [](f32 val) -> f32 {
+        return (val >= 0.0f) ? 1.0f : (val <= -5.3f) ? 0.0f : std::pow(10.0f, val);
+    };
+    const auto sin = [](f32 degrees) -> f32 {
+        return std::sin(degrees * std::numbers::pi_v<f32> / 180.0f);
+    };
+    const auto cos = [](f32 degrees) -> f32 {
+        return std::cos(degrees * std::numbers::pi_v<f32> / 180.0f);
+    };
+
+    Common::FixedPoint<50, 14> delay{static_cast<f32>(params.sample_rate) / 1000.0f};
+
+    Common::FixedPoint<50, 14> early_gain{
+        std::min(params.room_gain + params.reflection_gain, 5000.0f) / 2000.0f};
+    state.early_gain = pow_10(early_gain.to_float());
+    Common::FixedPoint<50, 14> late_gain{std::min(params.room_gain + params.reverb_gain, 5000.0f) /
+                                         2000.0f};
+    state.late_gain = pow_10(late_gain.to_float());
+
+    Common::FixedPoint<50, 14> hf_gain{pow_10(params.room_HF_gain / 2000.0f)};
+    if (hf_gain >= 1.0f) {
+        state.lowpass_1 = 0.0f;
+        state.lowpass_2 = 1.0f;
+    } else {
+        const auto reference_hf{(params.reference_HF * 256.0f) /
+                                static_cast<f32>(params.sample_rate)};
+        const Common::FixedPoint<50, 14> a{1.0f - hf_gain.to_float()};
+        const Common::FixedPoint<50, 14> b{2.0f + (-cos(reference_hf) * (hf_gain * 2.0f))};
+        const Common::FixedPoint<50, 14> c{
+            std::sqrt(std::pow(b.to_float(), 2.0f) + (std::pow(a.to_float(), 2.0f) * -4.0f))};
+
+        state.lowpass_1 = std::min(((b - c) / (a * 2.0f)).to_float(), 0.99723f);
+        state.lowpass_2 = 1.0f - state.lowpass_1;
+    }
+
+    state.early_to_late_taps =
+        (((params.reflection_delay + params.late_reverb_delay_time) * 1000.0f) * delay).to_int();
+    state.last_reverb_echo = params.late_reverb_diffusion * 0.6f * 0.01f;
+
+    for (u32 i = 0; i < I3dl2ReverbInfo::MaxDelayLines; i++) {
+        auto curr_delay{
+            ((MinDelayLineTimes[i] + (params.late_reverb_density / 100.0f) *
+                                         (MaxDelayLineTimes[i] - MinDelayLineTimes[i])) *
+             delay)
+                .to_int()};
+        state.fdn_delay_lines[i].SetDelay(curr_delay);
+
+        const auto a{
+            (static_cast<f32>(state.fdn_delay_lines[i].delay + state.decay_delay_lines0[i].delay +
+                              state.decay_delay_lines1[i].delay) *
+             -60.0f) /
+            (params.late_reverb_decay_time * static_cast<f32>(params.sample_rate))};
+        const auto b{a / params.late_reverb_HF_decay_ratio};
+        const auto c{
+            cos(((params.reference_HF * 0.5f) * 128.0f) / static_cast<f32>(params.sample_rate)) /
+            sin(((params.reference_HF * 0.5f) * 128.0f) / static_cast<f32>(params.sample_rate))};
+        const auto d{pow_10((b - a) / 40.0f)};
+        const auto e{pow_10((b + a) / 40.0f) * 0.7071f};
+
+        state.lowpass_coeff[i][0] = ((c * d + 1.0f) * e) / (c + d);
+        state.lowpass_coeff[i][1] = ((1.0f - (c * d)) * e) / (c + d);
+        state.lowpass_coeff[i][2] = (c - d) / (c + d);
+
+        state.decay_delay_lines0[i].wet_gain = state.last_reverb_echo;
+        state.decay_delay_lines1[i].wet_gain = state.last_reverb_echo * -0.9f;
+    }
+
+    if (reset) {
+        state.shelf_filter.fill(0.0f);
+        state.lowpass_0 = 0.0f;
+        for (u32 i = 0; i < I3dl2ReverbInfo::MaxDelayLines; i++) {
+            std::ranges::fill(state.fdn_delay_lines[i].buffer, 0);
+            std::ranges::fill(state.decay_delay_lines0[i].buffer, 0);
+            std::ranges::fill(state.decay_delay_lines1[i].buffer, 0);
+        }
+        std::ranges::fill(state.center_delay_line.buffer, 0);
+        std::ranges::fill(state.early_delay_line.buffer, 0);
+    }
+
+    const auto reflection_time{(params.late_reverb_delay_time * 0.9998f + 0.02f) * 1000.0f};
+    const auto reflection_delay{params.reflection_delay * 1000.0f};
+    for (u32 i = 0; i < I3dl2ReverbInfo::MaxDelayTaps; i++) {
+        auto length{((reflection_delay + reflection_time * EarlyTapTimes[i]) * delay).to_int()};
+        if (length >= state.early_delay_line.max_delay) {
+            length = state.early_delay_line.max_delay;
+        }
+        state.early_tap_steps[i] = length;
+    }
+}
+
+/**
+ * Initialize a new I3dl2ReverbInfo state according to the given parameters.
+ *
+ * @param params     - Input parameters to update the state.
+ * @param state      - State to be updated.
+ * @param workbuffer - Game-supplied memory for the state. (Unused)
+ */
+static void InitializeI3dl2ReverbEffect(const I3dl2ReverbInfo::ParameterVersion1& params,
+                                        I3dl2ReverbInfo::State& state, const CpuAddr workbuffer) {
+    state = {};
+    Common::FixedPoint<50, 14> delay{static_cast<f32>(params.sample_rate) / 1000};
+
+    for (u32 i = 0; i < I3dl2ReverbInfo::MaxDelayLines; i++) {
+        auto fdn_delay_time{(MaxDelayLineTimes[i] * delay).to_uint_floor()};
+        state.fdn_delay_lines[i].Initialize(fdn_delay_time);
+
+        auto decay0_delay_time{(Decay0MaxDelayLineTimes[i] * delay).to_uint_floor()};
+        state.decay_delay_lines0[i].Initialize(decay0_delay_time);
+
+        auto decay1_delay_time{(Decay1MaxDelayLineTimes[i] * delay).to_uint_floor()};
+        state.decay_delay_lines1[i].Initialize(decay1_delay_time);
+    }
+
+    const auto center_delay_time{(5 * delay).to_uint_floor()};
+    state.center_delay_line.Initialize(center_delay_time);
+
+    const auto early_delay_time{(400 * delay).to_uint_floor()};
+    state.early_delay_line.Initialize(early_delay_time);
+
+    UpdateI3dl2ReverbEffectParameter(params, state, true);
+}
+
+/**
+ * Pass-through the effect, copying input to output directly, with no reverb applied.
+ *
+ * @param inputs        - Array of input mix buffers to copy.
+ * @param outputs       - Array of output mix buffers to receive copy.
+ * @param channel_count - Number of channels in inputs and outputs.
+ * @param sample_count  - Number of samples within each channel (unused).
+ */
+static void ApplyI3dl2ReverbEffectBypass(std::span<std::span<const s32>> inputs,
+                                         std::span<std::span<s32>> outputs, const u32 channel_count,
+                                         [[maybe_unused]] const u32 sample_count) {
+    for (u32 i = 0; i < channel_count; i++) {
+        if (inputs[i].data() != outputs[i].data()) {
+            std::memcpy(outputs[i].data(), inputs[i].data(), outputs[i].size_bytes());
+        }
+    }
+}
+
+/**
+ * Tick the delay lines, reading and returning their current output, and writing a new decaying
+ * sample (mix).
+ *
+ * @param decay0 - The first decay line.
+ * @param decay1 - The second decay line.
+ * @param fdn    - Feedback delay network.
+ * @param mix    - The new calculated sample to be written and decayed.
+ * @return The next delayed and decayed sample.
+ */
+static Common::FixedPoint<50, 14> Axfx2AllPassTick(I3dl2ReverbInfo::I3dl2DelayLine& decay0,
+                                                   I3dl2ReverbInfo::I3dl2DelayLine& decay1,
+                                                   I3dl2ReverbInfo::I3dl2DelayLine& fdn,
+                                                   const Common::FixedPoint<50, 14> mix) {
+    auto val{decay0.Read()};
+    auto mixed{mix - (val * decay0.wet_gain)};
+    auto out{decay0.Tick(mixed) + (mixed * decay0.wet_gain)};
+
+    val = decay1.Read();
+    mixed = out - (val * decay1.wet_gain);
+    out = decay1.Tick(mixed) + (mixed * decay1.wet_gain);
+
+    fdn.Tick(out);
+    return out;
+}
+
+/**
+ * Impl. Apply a I3DL2 reverb according to the current state, on the input mix buffers,
+ * saving the results to the output mix buffers.
+ *
+ * @tparam Channels    - Number of channels to process. 1-6.
+                         Inputs/outputs should have this many buffers.
+ * @param state        - State to use, must be initialized (see InitializeI3dl2ReverbEffect).
+ * @param inputs       - Input mix buffers to perform the reverb on.
+ * @param outputs      - Output mix buffers to receive the reverbed samples.
+ * @param sample_count - Number of samples to process.
+ */
+template <size_t Channels>
+static void ApplyI3dl2ReverbEffect(I3dl2ReverbInfo::State& state,
+                                   std::span<std::span<const s32>> inputs,
+                                   std::span<std::span<s32>> outputs, const u32 sample_count) {
+    constexpr std::array<u8, I3dl2ReverbInfo::MaxDelayTaps> OutTapIndexes1Ch{
+        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    };
+    constexpr std::array<u8, I3dl2ReverbInfo::MaxDelayTaps> OutTapIndexes2Ch{
+        0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1,
+    };
+    constexpr std::array<u8, I3dl2ReverbInfo::MaxDelayTaps> OutTapIndexes4Ch{
+        0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 1, 1, 1, 0, 0, 0, 0, 3, 3, 3,
+    };
+    constexpr std::array<u8, I3dl2ReverbInfo::MaxDelayTaps> OutTapIndexes6Ch{
+        4, 0, 0, 1, 1, 1, 1, 2, 2, 2, 1, 1, 1, 0, 0, 0, 0, 2, 2, 2,
+    };
+
+    std::span<const u8> tap_indexes{};
+    if constexpr (Channels == 1) {
+        tap_indexes = OutTapIndexes1Ch;
+    } else if constexpr (Channels == 2) {
+        tap_indexes = OutTapIndexes2Ch;
+    } else if constexpr (Channels == 4) {
+        tap_indexes = OutTapIndexes4Ch;
+    } else if constexpr (Channels == 6) {
+        tap_indexes = OutTapIndexes6Ch;
+    }
+
+    for (u32 i = 0; i < sample_count; i++) {
+        Common::FixedPoint<50, 14> early_to_late_tap{
+            state.early_delay_line.TapOut(state.early_to_late_taps)};
+        std::array<Common::FixedPoint<50, 14>, Channels> output_samples{};
+
+        for (u32 early_tap = 0; early_tap < I3dl2ReverbInfo::MaxDelayTaps; early_tap++) {
+            output_samples[tap_indexes[early_tap]] +=
+                state.early_delay_line.TapOut(state.early_tap_steps[early_tap]) *
+                EarlyGains[early_tap];
+            if constexpr (Channels == 6) {
+                output_samples[5] +=
+                    state.early_delay_line.TapOut(state.early_tap_steps[early_tap]) *
+                    EarlyGains[early_tap];
+            }
+        }
+
+        Common::FixedPoint<50, 14> current_sample{};
+        for (u32 channel = 0; channel < Channels; channel++) {
+            current_sample += inputs[channel][i];
+        }
+
+        state.lowpass_0 =
+            (current_sample * state.lowpass_2 + state.lowpass_0 * state.lowpass_1).to_float();
+        state.early_delay_line.Tick(state.lowpass_0);
+
+        for (u32 channel = 0; channel < Channels; channel++) {
+            output_samples[channel] *= state.early_gain;
+        }
+
+        std::array<Common::FixedPoint<50, 14>, I3dl2ReverbInfo::MaxDelayLines> filtered_samples{};
+        for (u32 delay_line = 0; delay_line < I3dl2ReverbInfo::MaxDelayLines; delay_line++) {
+            filtered_samples[delay_line] =
+                state.fdn_delay_lines[delay_line].Read() * state.lowpass_coeff[delay_line][0] +
+                state.shelf_filter[delay_line];
+            state.shelf_filter[delay_line] =
+                (filtered_samples[delay_line] * state.lowpass_coeff[delay_line][2] +
+                 state.fdn_delay_lines[delay_line].Read() * state.lowpass_coeff[delay_line][1])
+                    .to_float();
+        }
+
+        const std::array<Common::FixedPoint<50, 14>, I3dl2ReverbInfo::MaxDelayLines> mix_matrix{
+            filtered_samples[1] + filtered_samples[2] + early_to_late_tap * state.late_gain,
+            -filtered_samples[0] - filtered_samples[3] + early_to_late_tap * state.late_gain,
+            filtered_samples[0] - filtered_samples[3] + early_to_late_tap * state.late_gain,
+            filtered_samples[1] - filtered_samples[2] + early_to_late_tap * state.late_gain,
+        };
+
+        std::array<Common::FixedPoint<50, 14>, I3dl2ReverbInfo::MaxDelayLines> allpass_samples{};
+        for (u32 delay_line = 0; delay_line < I3dl2ReverbInfo::MaxDelayLines; delay_line++) {
+            allpass_samples[delay_line] = Axfx2AllPassTick(
+                state.decay_delay_lines0[delay_line], state.decay_delay_lines1[delay_line],
+                state.fdn_delay_lines[delay_line], mix_matrix[delay_line]);
+        }
+
+        const auto out_channels{std::min(Channels, size_t(4))};
+        for (u32 channel = 0; channel < out_channels; channel++) {
+            auto out_sample{output_samples[channel] + allpass_samples[channel] +
+                            state.dry_gain * static_cast<f32>(inputs[channel][i])};
+            outputs[channel][i] =
+                static_cast<s32>(std::clamp(out_sample.to_float(), -8388600.0f, 8388600.0f));
+        }
+
+        if constexpr (Channels == 6) {
+            auto center{
+                state.center_delay_line.Tick((allpass_samples[2] - allpass_samples[3]) * 0.5f)};
+            auto out_sample{static_cast<f32>(inputs[4][i]) * state.dry_gain +
+                            output_samples[4] * state.early_gain + center};
+            outputs[4][i] =
+                static_cast<s32>(std::clamp(out_sample.to_float(), -8388600.0f, 8388600.0f));
+
+            out_sample = static_cast<f32>(inputs[5][i]) * state.dry_gain +
+                         output_samples[5] * state.early_gain + allpass_samples[3];
+            outputs[5][i] =
+                static_cast<s32>(std::clamp(out_sample.to_float(), -8388600.0f, 8388600.0f));
+        }
+    }
+}
+
+/**
+ * Apply a I3DL2 reverb if enabled, according to the current state, on the input mix buffers,
+ * saving the results to the output mix buffers.
+ *
+ * @param params       - Input parameters to use.
+ * @param state        - State to use, must be initialized (see InitializeI3dl2ReverbEffect).
+ * @param enabled      - If enabled, delay will be applied, otherwise input is copied to output.
+ * @param inputs       - Input mix buffers to performan the delay on.
+ * @param outputs      - Output mix buffers to receive the delayed samples.
+ * @param sample_count - Number of samples to process.
+ */
+static void ApplyI3dl2ReverbEffect(const I3dl2ReverbInfo::ParameterVersion1& params,
+                                   I3dl2ReverbInfo::State& state, const bool enabled,
+                                   std::span<std::span<const s32>> inputs,
+                                   std::span<std::span<s32>> outputs, const u32 sample_count) {
+    if (enabled) {
+        switch (params.channel_count) {
+        case 0:
+            return;
+        case 1:
+            ApplyI3dl2ReverbEffect<1>(state, inputs, outputs, sample_count);
+            break;
+        case 2:
+            ApplyI3dl2ReverbEffect<2>(state, inputs, outputs, sample_count);
+            break;
+        case 4:
+            ApplyI3dl2ReverbEffect<4>(state, inputs, outputs, sample_count);
+            break;
+        case 6:
+            ApplyI3dl2ReverbEffect<6>(state, inputs, outputs, sample_count);
+            break;
+        default:
+            ApplyI3dl2ReverbEffectBypass(inputs, outputs, params.channel_count, sample_count);
+            break;
+        }
+    } else {
+        ApplyI3dl2ReverbEffectBypass(inputs, outputs, params.channel_count, sample_count);
+    }
+}
+
+void I3dl2ReverbCommand::Dump([[maybe_unused]] const ADSP::CommandListProcessor& processor,
+                              std::string& string) {
+    string += fmt::format("I3dl2ReverbCommand\n\tenabled {} \n\tinputs: ", effect_enabled);
+    for (u32 i = 0; i < parameter.channel_count; i++) {
+        string += fmt::format("{:02X}, ", inputs[i]);
+    }
+    string += "\n\toutputs: ";
+    for (u32 i = 0; i < parameter.channel_count; i++) {
+        string += fmt::format("{:02X}, ", outputs[i]);
+    }
+    string += "\n";
+}
+
+void I3dl2ReverbCommand::Process(const ADSP::CommandListProcessor& processor) {
+    std::vector<std::span<const s32>> input_buffers(parameter.channel_count);
+    std::vector<std::span<s32>> output_buffers(parameter.channel_count);
+
+    for (u32 i = 0; i < parameter.channel_count; i++) {
+        input_buffers[i] = processor.mix_buffers.subspan(inputs[i] * processor.sample_count,
+                                                         processor.sample_count);
+        output_buffers[i] = processor.mix_buffers.subspan(outputs[i] * processor.sample_count,
+                                                          processor.sample_count);
+    }
+
+    auto state_{reinterpret_cast<I3dl2ReverbInfo::State*>(state)};
+
+    if (effect_enabled) {
+        if (parameter.state == I3dl2ReverbInfo::ParameterState::Updating) {
+            UpdateI3dl2ReverbEffectParameter(parameter, *state_, false);
+        } else if (parameter.state == I3dl2ReverbInfo::ParameterState::Initialized) {
+            InitializeI3dl2ReverbEffect(parameter, *state_, workbuffer);
+        }
+    }
+    ApplyI3dl2ReverbEffect(parameter, *state_, effect_enabled, input_buffers, output_buffers,
+                           processor.sample_count);
+}
+
+bool I3dl2ReverbCommand::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/i3dl2_reverb.h

@@ -0,0 +1,60 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <string>
+
+#include "audio_core/renderer/command/icommand.h"
+#include "audio_core/renderer/effect/effect_i3dl2_info.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command for a I3DL2Reverb effect. Apply a reverb to inputs mix buffer according to
+ * the I3DL2 spec, outputs receives the results.
+ */
+struct I3dl2ReverbCommand : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Input mix buffer offsets for each channel
+    std::array<s16, MaxChannels> inputs;
+    /// Output mix buffer offsets for each channel
+    std::array<s16, MaxChannels> outputs;
+    /// Input parameters
+    I3dl2ReverbInfo::ParameterVersion1 parameter;
+    /// State, updated each call
+    CpuAddr state;
+    /// Game-supplied workbuffer (Unused)
+    CpuAddr workbuffer;
+    /// Is this effect enabled?
+    bool effect_enabled;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/light_limiter.cpp

@@ -0,0 +1,202 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/effect/light_limiter.h"
+
+namespace AudioCore::AudioRenderer {
+/**
+ * Update the LightLimiterInfo state according to the given parameters.
+ * A no-op.
+ *
+ * @param params - Input parameters to update the state.
+ * @param state  - State to be updated.
+ */
+static void UpdateLightLimiterEffectParameter(const LightLimiterInfo::ParameterVersion2& params,
+                                              LightLimiterInfo::State& state) {}
+
+/**
+ * Initialize a new LightLimiterInfo state according to the given parameters.
+ *
+ * @param params     - Input parameters to update the state.
+ * @param state      - State to be updated.
+ * @param workbuffer - Game-supplied memory for the state. (Unused)
+ */
+static void InitializeLightLimiterEffect(const LightLimiterInfo::ParameterVersion2& params,
+                                         LightLimiterInfo::State& state, const CpuAddr workbuffer) {
+    state = {};
+    state.samples_average.fill(0.0f);
+    state.compression_gain.fill(1.0f);
+    state.look_ahead_sample_offsets.fill(0);
+    for (u32 i = 0; i < params.channel_count; i++) {
+        state.look_ahead_sample_buffers[i].resize(params.look_ahead_samples_max, 0.0f);
+    }
+}
+
+/**
+ * Apply a light limiter effect if enabled, according to the current state, on the input mix
+ * buffers, saving the results to the output mix buffers.
+ *
+ * @param params       - Input parameters to use.
+ * @param state        - State to use, must be initialized (see InitializeLightLimiterEffect).
+ * @param enabled      - If enabled, limiter will be applied, otherwise input is copied to output.
+ * @param inputs       - Input mix buffers to perform the limiter on.
+ * @param outputs      - Output mix buffers to receive the limited samples.
+ * @param sample_count - Number of samples to process.
+ * @params statistics  - Optional output statistics, only used with version 2.
+ */
+static void ApplyLightLimiterEffect(const LightLimiterInfo::ParameterVersion2& params,
+                                    LightLimiterInfo::State& state, const bool enabled,
+                                    std::vector<std::span<const s32>>& inputs,
+                                    std::vector<std::span<s32>>& outputs, const u32 sample_count,
+                                    LightLimiterInfo::StatisticsInternal* statistics) {
+    constexpr s64 min{std::numeric_limits<s32>::min()};
+    constexpr s64 max{std::numeric_limits<s32>::max()};
+
+    if (enabled) {
+        if (statistics && params.statistics_reset_required) {
+            for (u32 i = 0; i < params.channel_count; i++) {
+                statistics->channel_compression_gain_min[i] = 1.0f;
+                statistics->channel_max_sample[i] = 0;
+            }
+        }
+
+        for (u32 sample_index = 0; sample_index < sample_count; sample_index++) {
+            for (u32 channel = 0; channel < params.channel_count; channel++) {
+                auto sample{Common::FixedPoint<49, 15>(inputs[channel][sample_index]) *
+                            params.input_gain};
+                auto abs_sample{sample};
+                if (sample < 0.0f) {
+                    abs_sample = -sample;
+                }
+                auto coeff{abs_sample > state.samples_average[channel] ? params.attack_coeff
+                                                                       : params.release_coeff};
+                state.samples_average[channel] +=
+                    ((abs_sample - state.samples_average[channel]) * coeff).to_float();
+
+                auto above_threshold{state.samples_average[channel] > params.threshold};
+                auto attenuation{above_threshold ? params.threshold / state.samples_average[channel]
+                                                 : 1.0f};
+                coeff = attenuation < state.compression_gain[channel] ? params.attack_coeff
+                                                                      : params.release_coeff;
+                state.compression_gain[channel] +=
+                    (attenuation - state.compression_gain[channel]) * coeff;
+
+                auto lookahead_sample{
+                    state.look_ahead_sample_buffers[channel]
+                                                   [state.look_ahead_sample_offsets[channel]]};
+                state.look_ahead_sample_buffers[channel][state.look_ahead_sample_offsets[channel]] =
+                    sample;
+                state.look_ahead_sample_offsets[channel] =
+                    (state.look_ahead_sample_offsets[channel] + 1) % params.look_ahead_samples_min;
+
+                outputs[channel][sample_index] = static_cast<s32>(std::clamp(
+                    (lookahead_sample * state.compression_gain[channel] * params.output_gain)
+                        .to_long(),
+                    min, max));
+
+                if (statistics) {
+                    statistics->channel_max_sample[channel] =
+                        std::max(statistics->channel_max_sample[channel], abs_sample.to_float());
+                    statistics->channel_compression_gain_min[channel] =
+                        std::min(statistics->channel_compression_gain_min[channel],
+                                 state.compression_gain[channel].to_float());
+                }
+            }
+        }
+    } else {
+        for (u32 i = 0; i < params.channel_count; i++) {
+            if (params.inputs[i] != params.outputs[i]) {
+                std::memcpy(outputs[i].data(), inputs[i].data(), outputs[i].size_bytes());
+            }
+        }
+    }
+}
+
+void LightLimiterVersion1Command::Dump([[maybe_unused]] const ADSP::CommandListProcessor& processor,
+                                       std::string& string) {
+    string += fmt::format("LightLimiterVersion1Command\n\tinputs: ");
+    for (u32 i = 0; i < MaxChannels; i++) {
+        string += fmt::format("{:02X}, ", inputs[i]);
+    }
+    string += "\n\toutputs: ";
+    for (u32 i = 0; i < MaxChannels; i++) {
+        string += fmt::format("{:02X}, ", outputs[i]);
+    }
+    string += "\n";
+}
+
+void LightLimiterVersion1Command::Process(const ADSP::CommandListProcessor& processor) {
+    std::vector<std::span<const s32>> input_buffers(parameter.channel_count);
+    std::vector<std::span<s32>> output_buffers(parameter.channel_count);
+
+    for (u32 i = 0; i < parameter.channel_count; i++) {
+        input_buffers[i] = processor.mix_buffers.subspan(inputs[i] * processor.sample_count,
+                                                         processor.sample_count);
+        output_buffers[i] = processor.mix_buffers.subspan(outputs[i] * processor.sample_count,
+                                                          processor.sample_count);
+    }
+
+    auto state_{reinterpret_cast<LightLimiterInfo::State*>(state)};
+
+    if (effect_enabled) {
+        if (parameter.state == LightLimiterInfo::ParameterState::Updating) {
+            UpdateLightLimiterEffectParameter(parameter, *state_);
+        } else if (parameter.state == LightLimiterInfo::ParameterState::Initialized) {
+            InitializeLightLimiterEffect(parameter, *state_, workbuffer);
+        }
+    }
+
+    LightLimiterInfo::StatisticsInternal* statistics{nullptr};
+    ApplyLightLimiterEffect(parameter, *state_, effect_enabled, input_buffers, output_buffers,
+                            processor.sample_count, statistics);
+}
+
+bool LightLimiterVersion1Command::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+void LightLimiterVersion2Command::Dump([[maybe_unused]] const ADSP::CommandListProcessor& processor,
+                                       std::string& string) {
+    string += fmt::format("LightLimiterVersion2Command\n\tinputs: \n");
+    for (u32 i = 0; i < MaxChannels; i++) {
+        string += fmt::format("{:02X}, ", inputs[i]);
+    }
+    string += "\n\toutputs: ";
+    for (u32 i = 0; i < MaxChannels; i++) {
+        string += fmt::format("{:02X}, ", outputs[i]);
+    }
+    string += "\n";
+}
+
+void LightLimiterVersion2Command::Process(const ADSP::CommandListProcessor& processor) {
+    auto state_{reinterpret_cast<LightLimiterInfo::State*>(state)};
+
+    std::vector<std::span<const s32>> input_buffers(parameter.channel_count);
+    std::vector<std::span<s32>> output_buffers(parameter.channel_count);
+
+    for (u32 i = 0; i < parameter.channel_count; i++) {
+        input_buffers[i] = processor.mix_buffers.subspan(inputs[i] * processor.sample_count,
+                                                         processor.sample_count);
+        output_buffers[i] = processor.mix_buffers.subspan(outputs[i] * processor.sample_count,
+                                                          processor.sample_count);
+    }
+
+    if (effect_enabled) {
+        if (parameter.state == LightLimiterInfo::ParameterState::Updating) {
+            UpdateLightLimiterEffectParameter(parameter, *state_);
+        } else if (parameter.state == LightLimiterInfo::ParameterState::Initialized) {
+            InitializeLightLimiterEffect(parameter, *state_, workbuffer);
+        }
+    }
+
+    auto statistics{reinterpret_cast<LightLimiterInfo::StatisticsInternal*>(result_state)};
+    ApplyLightLimiterEffect(parameter, *state_, effect_enabled, input_buffers, output_buffers,
+                            processor.sample_count, statistics);
+}
+
+bool LightLimiterVersion2Command::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/light_limiter.h

@@ -0,0 +1,103 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <string>
+
+#include "audio_core/renderer/command/icommand.h"
+#include "audio_core/renderer/effect/effect_light_limiter_info.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command for limiting volume between a high and low threshold.
+ * Version 1.
+ */
+struct LightLimiterVersion1Command : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Input mix buffer offsets for each channel
+    std::array<s16, MaxChannels> inputs;
+    /// Output mix buffer offsets for each channel
+    std::array<s16, MaxChannels> outputs;
+    /// Input parameters
+    LightLimiterInfo::ParameterVersion2 parameter;
+    /// State, updated each call
+    CpuAddr state;
+    /// Game-supplied workbuffer (Unused)
+    CpuAddr workbuffer;
+    /// Is this effect enabled?
+    bool effect_enabled;
+};
+
+/**
+ * AudioRenderer command for limiting volume between a high and low threshold.
+ * Version 2 with output statistics.
+ */
+struct LightLimiterVersion2Command : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Input mix buffer offsets for each channel
+    std::array<s16, MaxChannels> inputs;
+    /// Output mix buffer offsets for each channel
+    std::array<s16, MaxChannels> outputs;
+    /// Input parameters
+    LightLimiterInfo::ParameterVersion2 parameter;
+    /// State, updated each call
+    CpuAddr state;
+    /// Game-supplied workbuffer (Unused)
+    CpuAddr workbuffer;
+    /// Optional statistics, sent back to the sysmodule
+    CpuAddr result_state;
+    /// Is this effect enabled?
+    bool effect_enabled;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/multi_tap_biquad_filter.cpp

@@ -0,0 +1,45 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/effect/biquad_filter.h"
+#include "audio_core/renderer/command/effect/multi_tap_biquad_filter.h"
+
+namespace AudioCore::AudioRenderer {
+
+void MultiTapBiquadFilterCommand::Dump([[maybe_unused]] const ADSP::CommandListProcessor& processor,
+                                       std::string& string) {
+    string += fmt::format(
+        "MultiTapBiquadFilterCommand\n\tinput {:02X}\n\toutput {:02X}\n\tneeds_init ({}, {})\n",
+        input, output, needs_init[0], needs_init[1]);
+}
+
+void MultiTapBiquadFilterCommand::Process(const ADSP::CommandListProcessor& processor) {
+    if (filter_tap_count > MaxBiquadFilters) {
+        LOG_ERROR(Service_Audio, "Too many filter taps! {}", filter_tap_count);
+        filter_tap_count = MaxBiquadFilters;
+    }
+
+    auto input_buffer{
+        processor.mix_buffers.subspan(input * processor.sample_count, processor.sample_count)};
+    auto output_buffer{
+        processor.mix_buffers.subspan(output * processor.sample_count, processor.sample_count)};
+
+    // TODO: Fix this, currently just applies the filter to the input twice,
+    // and doesn't chain the biquads together at all.
+    for (u32 i = 0; i < filter_tap_count; i++) {
+        auto state{reinterpret_cast<VoiceState::BiquadFilterState*>(states[i])};
+        if (needs_init[i]) {
+            std::memset(state, 0, sizeof(VoiceState::BiquadFilterState));
+        }
+
+        ApplyBiquadFilterFloat(output_buffer, input_buffer, biquads[i].b, biquads[i].a, *state,
+                               processor.sample_count);
+    }
+}
+
+bool MultiTapBiquadFilterCommand::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/multi_tap_biquad_filter.h

@@ -0,0 +1,59 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <string>
+
+#include "audio_core/renderer/command/icommand.h"
+#include "audio_core/renderer/voice/voice_info.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command for applying multiple biquads at once.
+ */
+struct MultiTapBiquadFilterCommand : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Input mix buffer index
+    s16 input;
+    /// Output mix buffer index
+    s16 output;
+    /// Biquad parameters
+    std::array<VoiceInfo::BiquadFilterParameter, MaxBiquadFilters> biquads;
+    /// Biquad states, updated each call
+    std::array<CpuAddr, MaxBiquadFilters> states;
+    /// If each biquad needs initialisation
+    std::array<bool, MaxBiquadFilters> needs_init;
+    /// Number of active biquads
+    u8 filter_tap_count;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/reverb.cpp

@@ -0,0 +1,433 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <numbers>
+#include <ranges>
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/effect/reverb.h"
+
+namespace AudioCore::AudioRenderer {
+
+constexpr std::array<f32, ReverbInfo::MaxDelayLines> FdnMaxDelayLineTimes = {
+    53.9532470703125f,
+    79.19256591796875f,
+    116.23876953125f,
+    170.61529541015625f,
+};
+
+constexpr std::array<f32, ReverbInfo::MaxDelayLines> DecayMaxDelayLineTimes = {
+    7.0f,
+    9.0f,
+    13.0f,
+    17.0f,
+};
+
+constexpr std::array<std::array<f32, ReverbInfo::MaxDelayTaps + 1>, ReverbInfo::NumEarlyModes>
+    EarlyDelayTimes = {
+        {{0.000000f, 3.500000f, 2.799988f, 3.899963f, 2.699951f, 13.399963f, 7.899963f, 8.399963f,
+          9.899963f, 12.000000f, 12.500000f},
+         {0.000000f, 11.799988f, 5.500000f, 11.199951f, 10.399963f, 38.099976f, 22.199951f,
+          29.599976f, 21.199951f, 24.799988f, 40.000000f},
+         {0.000000f, 41.500000f, 20.500000f, 41.299988f, 0.000000f, 29.500000f, 33.799988f,
+          45.199951f, 46.799988f, 0.000000f, 50.000000f},
+         {33.099976f, 43.299988f, 22.799988f, 37.899963f, 14.899963f, 35.299988f, 17.899963f,
+          34.199951f, 0.000000f, 43.299988f, 50.000000f},
+         {0.000000f, 0.000000f, 0.000000f, 0.000000f, 0.000000f, 0.000000f, 0.000000f, 0.000000f,
+          0.000000f, 0.000000f, 0.000000f}},
+};
+
+constexpr std::array<std::array<f32, ReverbInfo::MaxDelayTaps>, ReverbInfo::NumEarlyModes>
+    EarlyDelayGains = {{
+        {0.699951f, 0.679993f, 0.699951f, 0.679993f, 0.699951f, 0.679993f, 0.699951f, 0.679993f,
+         0.679993f, 0.679993f},
+        {0.699951f, 0.679993f, 0.699951f, 0.679993f, 0.699951f, 0.679993f, 0.679993f, 0.679993f,
+         0.679993f, 0.679993f},
+        {0.500000f, 0.699951f, 0.699951f, 0.679993f, 0.500000f, 0.679993f, 0.679993f, 0.699951f,
+         0.679993f, 0.000000f},
+        {0.929993f, 0.919983f, 0.869995f, 0.859985f, 0.939941f, 0.809998f, 0.799988f, 0.769958f,
+         0.759949f, 0.649963f},
+        {0.000000f, 0.000000f, 0.000000f, 0.000000f, 0.000000f, 0.000000f, 0.000000f, 0.000000f,
+         0.000000f, 0.000000f},
+    }};
+
+constexpr std::array<std::array<f32, ReverbInfo::MaxDelayLines>, ReverbInfo::NumLateModes>
+    FdnDelayTimes = {{
+        {53.953247f, 79.192566f, 116.238770f, 130.615295f},
+        {53.953247f, 79.192566f, 116.238770f, 170.615295f},
+        {5.000000f, 10.000000f, 5.000000f, 10.000000f},
+        {47.029968f, 71.000000f, 103.000000f, 170.000000f},
+        {53.953247f, 79.192566f, 116.238770f, 170.615295f},
+    }};
+
+constexpr std::array<std::array<f32, ReverbInfo::MaxDelayLines>, ReverbInfo::NumLateModes>
+    DecayDelayTimes = {{
+        {7.000000f, 9.000000f, 13.000000f, 17.000000f},
+        {7.000000f, 9.000000f, 13.000000f, 17.000000f},
+        {1.000000f, 1.000000f, 1.000000f, 1.000000f},
+        {7.000000f, 7.000000f, 13.000000f, 9.000000f},
+        {7.000000f, 9.000000f, 13.000000f, 17.000000f},
+    }};
+
+/**
+ * Update the ReverbInfo state according to the given parameters.
+ *
+ * @param params - Input parameters to update the state.
+ * @param state  - State to be updated.
+ */
+static void UpdateReverbEffectParameter(const ReverbInfo::ParameterVersion2& params,
+                                        ReverbInfo::State& state) {
+    const auto pow_10 = [](f32 val) -> f32 {
+        return (val >= 0.0f) ? 1.0f : (val <= -5.3f) ? 0.0f : std::pow(10.0f, val);
+    };
+    const auto cos = [](f32 degrees) -> f32 {
+        return std::cos(degrees * std::numbers::pi_v<f32> / 180.0f);
+    };
+
+    static bool unk_initialized{false};
+    static Common::FixedPoint<50, 14> unk_value{};
+
+    const auto sample_rate{Common::FixedPoint<50, 14>::from_base(params.sample_rate)};
+    const auto pre_delay_time{Common::FixedPoint<50, 14>::from_base(params.pre_delay)};
+
+    for (u32 i = 0; i < ReverbInfo::MaxDelayTaps; i++) {
+        auto early_delay{
+            ((pre_delay_time + EarlyDelayTimes[params.early_mode][i]) * sample_rate).to_int()};
+        early_delay = std::min(early_delay, state.pre_delay_line.sample_count_max);
+        state.early_delay_times[i] = early_delay + 1;
+        state.early_gains[i] = Common::FixedPoint<50, 14>::from_base(params.early_gain) *
+                               EarlyDelayGains[params.early_mode][i];
+    }
+
+    if (params.channel_count == 2) {
+        state.early_gains[4] * 0.5f;
+        state.early_gains[5] * 0.5f;
+    }
+
+    auto pre_time{
+        ((pre_delay_time + EarlyDelayTimes[params.early_mode][10]) * sample_rate).to_int()};
+    state.pre_delay_time = std::min(pre_time, state.pre_delay_line.sample_count_max);
+
+    if (!unk_initialized) {
+        unk_value = cos((1280.0f / sample_rate).to_float());
+        unk_initialized = true;
+    }
+
+    for (u32 i = 0; i < ReverbInfo::MaxDelayLines; i++) {
+        const auto fdn_delay{(FdnDelayTimes[params.late_mode][i] * sample_rate).to_int()};
+        state.fdn_delay_lines[i].sample_count =
+            std::min(fdn_delay, state.fdn_delay_lines[i].sample_count_max);
+        state.fdn_delay_lines[i].buffer_end =
+            &state.fdn_delay_lines[i].buffer[state.fdn_delay_lines[i].sample_count - 1];
+
+        const auto decay_delay{(DecayDelayTimes[params.late_mode][i] * sample_rate).to_int()};
+        state.decay_delay_lines[i].sample_count =
+            std::min(decay_delay, state.decay_delay_lines[i].sample_count_max);
+        state.decay_delay_lines[i].buffer_end =
+            &state.decay_delay_lines[i].buffer[state.decay_delay_lines[i].sample_count - 1];
+
+        state.decay_delay_lines[i].decay =
+            0.5999755859375f * (1.0f - Common::FixedPoint<50, 14>::from_base(params.colouration));
+
+        auto a{(Common::FixedPoint<50, 14>(state.fdn_delay_lines[i].sample_count_max) +
+                state.decay_delay_lines[i].sample_count_max) *
+               -3};
+        auto b{a / (Common::FixedPoint<50, 14>::from_base(params.decay_time) * sample_rate)};
+        Common::FixedPoint<50, 14> c{0.0f};
+        Common::FixedPoint<50, 14> d{0.0f};
+        auto hf_decay_ratio{Common::FixedPoint<50, 14>::from_base(params.high_freq_decay_ratio)};
+
+        if (hf_decay_ratio > 0.99493408203125f) {
+            c = 0.0f;
+            d = 1.0f;
+        } else {
+            const auto e{
+                pow_10(((((1.0f / hf_decay_ratio) - 1.0f) * 2) / 100 * (b / 10)).to_float())};
+            const auto f{1.0f - e};
+            const auto g{2.0f - (unk_value * e * 2)};
+            const auto h{std::sqrt(std::pow(g.to_float(), 2.0f) - (std::pow(f, 2.0f) * 4))};
+
+            c = (g - h) / (f * 2.0f);
+            d = 1.0f - c;
+        }
+
+        state.hf_decay_prev_gain[i] = c;
+        state.hf_decay_gain[i] = pow_10((b / 1000).to_float()) * d * 0.70709228515625f;
+        state.prev_feedback_output[i] = 0;
+    }
+}
+
+/**
+ * Initialize a new ReverbInfo state according to the given parameters.
+ *
+ * @param params                        - Input parameters to update the state.
+ * @param state                         - State to be updated.
+ * @param workbuffer                    - Game-supplied memory for the state. (Unused)
+ * @param long_size_pre_delay_supported - Use a longer pre-delay time before reverb begins.
+ */
+static void InitializeReverbEffect(const ReverbInfo::ParameterVersion2& params,
+                                   ReverbInfo::State& state, const CpuAddr workbuffer,
+                                   const bool long_size_pre_delay_supported) {
+    state = {};
+
+    auto delay{Common::FixedPoint<50, 14>::from_base(params.sample_rate)};
+
+    for (u32 i = 0; i < ReverbInfo::MaxDelayLines; i++) {
+        auto fdn_delay_time{(FdnMaxDelayLineTimes[i] * delay).to_uint_floor()};
+        state.fdn_delay_lines[i].Initialize(fdn_delay_time, 1.0f);
+
+        auto decay_delay_time{(DecayMaxDelayLineTimes[i] * delay).to_uint_floor()};
+        state.decay_delay_lines[i].Initialize(decay_delay_time, 0.0f);
+    }
+
+    const auto pre_delay{long_size_pre_delay_supported ? 350.0f : 150.0f};
+    const auto pre_delay_line{(pre_delay * delay).to_uint_floor()};
+    state.pre_delay_line.Initialize(pre_delay_line, 1.0f);
+
+    const auto center_delay_time{(5 * delay).to_uint_floor()};
+    state.center_delay_line.Initialize(center_delay_time, 1.0f);
+
+    UpdateReverbEffectParameter(params, state);
+
+    for (u32 i = 0; i < ReverbInfo::MaxDelayLines; i++) {
+        std::ranges::fill(state.fdn_delay_lines[i].buffer, 0);
+        std::ranges::fill(state.decay_delay_lines[i].buffer, 0);
+    }
+    std::ranges::fill(state.center_delay_line.buffer, 0);
+    std::ranges::fill(state.pre_delay_line.buffer, 0);
+}
+
+/**
+ * Pass-through the effect, copying input to output directly, with no reverb applied.
+ *
+ * @param inputs        - Array of input mix buffers to copy.
+ * @param outputs       - Array of output mix buffers to receive copy.
+ * @param channel_count - Number of channels in inputs and outputs.
+ * @param sample_count  - Number of samples within each channel.
+ */
+static void ApplyReverbEffectBypass(std::span<std::span<const s32>> inputs,
+                                    std::span<std::span<s32>> outputs, const u32 channel_count,
+                                    const u32 sample_count) {
+    for (u32 i = 0; i < channel_count; i++) {
+        if (inputs[i].data() != outputs[i].data()) {
+            std::memcpy(outputs[i].data(), inputs[i].data(), outputs[i].size_bytes());
+        }
+    }
+}
+
+/**
+ * Tick the delay lines, reading and returning their current output, and writing a new decaying
+ * sample (mix).
+ *
+ * @param decay  - The decay line.
+ * @param fdn    - Feedback delay network.
+ * @param mix    - The new calculated sample to be written and decayed.
+ * @return The next delayed and decayed sample.
+ */
+static Common::FixedPoint<50, 14> Axfx2AllPassTick(ReverbInfo::ReverbDelayLine& decay,
+                                                   ReverbInfo::ReverbDelayLine& fdn,
+                                                   const Common::FixedPoint<50, 14> mix) {
+    const auto val{decay.Read()};
+    const auto mixed{mix - (val * decay.decay)};
+    const auto out{decay.Tick(mixed) + (mixed * decay.decay)};
+
+    fdn.Tick(out);
+    return out;
+}
+
+/**
+ * Impl. Apply a Reverb according to the current state, on the input mix buffers,
+ * saving the results to the output mix buffers.
+ *
+ * @tparam Channels    - Number of channels to process. 1-6.
+                         Inputs/outputs should have this many buffers.
+ * @param params       - Input parameters to update the state.
+ * @param state        - State to use, must be initialized (see InitializeReverbEffect).
+ * @param inputs       - Input mix buffers to perform the reverb on.
+ * @param outputs      - Output mix buffers to receive the reverbed samples.
+ * @param sample_count - Number of samples to process.
+ */
+template <size_t Channels>
+static void ApplyReverbEffect(const ReverbInfo::ParameterVersion2& params, ReverbInfo::State& state,
+                              std::vector<std::span<const s32>>& inputs,
+                              std::vector<std::span<s32>>& outputs, const u32 sample_count) {
+    constexpr std::array<u8, ReverbInfo::MaxDelayTaps> OutTapIndexes1Ch{
+        0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    };
+    constexpr std::array<u8, ReverbInfo::MaxDelayTaps> OutTapIndexes2Ch{
+        0, 0, 1, 1, 0, 1, 0, 0, 1, 1,
+    };
+    constexpr std::array<u8, ReverbInfo::MaxDelayTaps> OutTapIndexes4Ch{
+        0, 0, 1, 1, 0, 1, 2, 2, 3, 3,
+    };
+    constexpr std::array<u8, ReverbInfo::MaxDelayTaps> OutTapIndexes6Ch{
+        0, 0, 1, 1, 4, 4, 2, 2, 3, 3,
+    };
+
+    std::span<const u8> tap_indexes{};
+    if constexpr (Channels == 1) {
+        tap_indexes = OutTapIndexes1Ch;
+    } else if constexpr (Channels == 2) {
+        tap_indexes = OutTapIndexes2Ch;
+    } else if constexpr (Channels == 4) {
+        tap_indexes = OutTapIndexes4Ch;
+    } else if constexpr (Channels == 6) {
+        tap_indexes = OutTapIndexes6Ch;
+    }
+
+    for (u32 sample_index = 0; sample_index < sample_count; sample_index++) {
+        std::array<Common::FixedPoint<50, 14>, Channels> output_samples{};
+
+        for (u32 early_tap = 0; early_tap < ReverbInfo::MaxDelayTaps; early_tap++) {
+            const auto sample{state.pre_delay_line.TapOut(state.early_delay_times[early_tap]) *
+                              state.early_gains[early_tap]};
+            output_samples[tap_indexes[early_tap]] += sample;
+            if constexpr (Channels == 6) {
+                output_samples[5] += sample;
+            }
+        }
+
+        if constexpr (Channels == 6) {
+            output_samples[5] *= 0.2f;
+        }
+
+        Common::FixedPoint<50, 14> input_sample{};
+        for (u32 channel = 0; channel < Channels; channel++) {
+            input_sample += inputs[channel][sample_index];
+        }
+
+        input_sample *= 64;
+        input_sample *= Common::FixedPoint<50, 14>::from_base(params.base_gain);
+        state.pre_delay_line.Write(input_sample);
+
+        for (u32 i = 0; i < ReverbInfo::MaxDelayLines; i++) {
+            state.prev_feedback_output[i] =
+                state.prev_feedback_output[i] * state.hf_decay_prev_gain[i] +
+                state.fdn_delay_lines[i].Read() * state.hf_decay_gain[i];
+        }
+
+        Common::FixedPoint<50, 14> pre_delay_sample{
+            state.pre_delay_line.Read() * Common::FixedPoint<50, 14>::from_base(params.late_gain)};
+
+        std::array<Common::FixedPoint<50, 14>, ReverbInfo::MaxDelayLines> mix_matrix{
+            state.prev_feedback_output[2] + state.prev_feedback_output[1] + pre_delay_sample,
+            -state.prev_feedback_output[0] - state.prev_feedback_output[3] + pre_delay_sample,
+            state.prev_feedback_output[0] - state.prev_feedback_output[3] + pre_delay_sample,
+            state.prev_feedback_output[1] - state.prev_feedback_output[2] + pre_delay_sample,
+        };
+
+        std::array<Common::FixedPoint<50, 14>, ReverbInfo::MaxDelayLines> out_line_samples{};
+        for (u32 i = 0; i < ReverbInfo::MaxDelayLines; i++) {
+            out_line_samples[i] = Axfx2AllPassTick(state.decay_delay_lines[i],
+                                                   state.fdn_delay_lines[i], mix_matrix[i]);
+        }
+        const auto dry_gain{Common::FixedPoint<50, 14>::from_base(params.dry_gain)};
+        const auto wet_gain{Common::FixedPoint<50, 14>::from_base(params.wet_gain)};
+
+        const auto out_channels{std::min(Channels, size_t(4))};
+        for (u32 channel = 0; channel < out_channels; channel++) {
+            auto in_sample{inputs[channel][channel] * dry_gain};
+            auto out_sample{((output_samples[channel] + out_line_samples[channel]) * wet_gain) /
+                            64};
+            outputs[channel][sample_index] = (in_sample + out_sample).to_int();
+        }
+
+        if constexpr (Channels == 6) {
+            auto center{
+                state.center_delay_line.Tick((out_line_samples[2] - out_line_samples[3]) * 0.5f)};
+            auto in_sample{inputs[4][sample_index] * dry_gain};
+            auto out_sample{((output_samples[4] + center) * wet_gain) / 64};
+
+            outputs[4][sample_index] = (in_sample + out_sample).to_int();
+
+            in_sample = inputs[5][sample_index] * dry_gain;
+            out_sample = ((output_samples[5] + out_line_samples[3]) * wet_gain) / 64;
+
+            outputs[5][sample_index] = (in_sample + out_sample).to_int();
+        }
+    }
+}
+
+/**
+ * Apply a Reverb if enabled, according to the current state, on the input mix buffers,
+ * saving the results to the output mix buffers.
+ *
+ * @param params       - Input parameters to use.
+ * @param state        - State to use, must be initialized (see InitializeReverbEffect).
+ * @param enabled      - If enabled, delay will be applied, otherwise input is copied to output.
+ * @param inputs       - Input mix buffers to performan the reverb on.
+ * @param outputs      - Output mix buffers to receive the reverbed samples.
+ * @param sample_count - Number of samples to process.
+ */
+static void ApplyReverbEffect(const ReverbInfo::ParameterVersion2& params, ReverbInfo::State& state,
+                              const bool enabled, std::vector<std::span<const s32>>& inputs,
+                              std::vector<std::span<s32>>& outputs, const u32 sample_count) {
+    if (enabled) {
+        switch (params.channel_count) {
+        case 0:
+            return;
+        case 1:
+            ApplyReverbEffect<1>(params, state, inputs, outputs, sample_count);
+            break;
+        case 2:
+            ApplyReverbEffect<2>(params, state, inputs, outputs, sample_count);
+            break;
+        case 4:
+            ApplyReverbEffect<4>(params, state, inputs, outputs, sample_count);
+            break;
+        case 6:
+            ApplyReverbEffect<6>(params, state, inputs, outputs, sample_count);
+            break;
+        default:
+            ApplyReverbEffectBypass(inputs, outputs, params.channel_count, sample_count);
+            break;
+        }
+    } else {
+        ApplyReverbEffectBypass(inputs, outputs, params.channel_count, sample_count);
+    }
+}
+
+void ReverbCommand::Dump([[maybe_unused]] const ADSP::CommandListProcessor& processor,
+                         std::string& string) {
+    string += fmt::format(
+        "ReverbCommand\n\tenabled {} long_size_pre_delay_supported {}\n\tinputs: ", effect_enabled,
+        long_size_pre_delay_supported);
+    for (u32 i = 0; i < MaxChannels; i++) {
+        string += fmt::format("{:02X}, ", inputs[i]);
+    }
+    string += "\n\toutputs: ";
+    for (u32 i = 0; i < MaxChannels; i++) {
+        string += fmt::format("{:02X}, ", outputs[i]);
+    }
+    string += "\n";
+}
+
+void ReverbCommand::Process(const ADSP::CommandListProcessor& processor) {
+    std::vector<std::span<const s32>> input_buffers(parameter.channel_count);
+    std::vector<std::span<s32>> output_buffers(parameter.channel_count);
+
+    for (u32 i = 0; i < parameter.channel_count; i++) {
+        input_buffers[i] = processor.mix_buffers.subspan(inputs[i] * processor.sample_count,
+                                                         processor.sample_count);
+        output_buffers[i] = processor.mix_buffers.subspan(outputs[i] * processor.sample_count,
+                                                          processor.sample_count);
+    }
+
+    auto state_{reinterpret_cast<ReverbInfo::State*>(state)};
+
+    if (effect_enabled) {
+        if (parameter.state == ReverbInfo::ParameterState::Updating) {
+            UpdateReverbEffectParameter(parameter, *state_);
+        } else if (parameter.state == ReverbInfo::ParameterState::Initialized) {
+            InitializeReverbEffect(parameter, *state_, workbuffer, long_size_pre_delay_supported);
+        }
+    }
+    ApplyReverbEffect(parameter, *state_, effect_enabled, input_buffers, output_buffers,
+                      processor.sample_count);
+}
+
+bool ReverbCommand::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/effect/reverb.h

@@ -0,0 +1,62 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <string>
+
+#include "audio_core/renderer/command/icommand.h"
+#include "audio_core/renderer/effect/effect_reverb_info.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command for a Reverb effect. Apply a reverb to inputs mix buffer, outputs receives
+ * the results.
+ */
+struct ReverbCommand : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Input mix buffer offsets for each channel
+    std::array<s16, MaxChannels> inputs;
+    /// Output mix buffer offsets for each channel
+    std::array<s16, MaxChannels> outputs;
+    /// Input parameters
+    ReverbInfo::ParameterVersion2 parameter;
+    /// State, updated each call
+    CpuAddr state;
+    /// Game-supplied workbuffer (Unused)
+    CpuAddr workbuffer;
+    /// Is this effect enabled?
+    bool effect_enabled;
+    /// Is a longer pre-delay time supported?
+    bool long_size_pre_delay_supported;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/icommand.h

@@ -0,0 +1,92 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "audio_core/common/common.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+enum class CommandId : u8 {
+    /* 0x00 */ Invalid,
+    /* 0x01 */ DataSourcePcmInt16Version1,
+    /* 0x02 */ DataSourcePcmInt16Version2,
+    /* 0x03 */ DataSourcePcmFloatVersion1,
+    /* 0x04 */ DataSourcePcmFloatVersion2,
+    /* 0x05 */ DataSourceAdpcmVersion1,
+    /* 0x06 */ DataSourceAdpcmVersion2,
+    /* 0x07 */ Volume,
+    /* 0x08 */ VolumeRamp,
+    /* 0x09 */ BiquadFilter,
+    /* 0x0A */ Mix,
+    /* 0x0B */ MixRamp,
+    /* 0x0C */ MixRampGrouped,
+    /* 0x0D */ DepopPrepare,
+    /* 0x0E */ DepopForMixBuffers,
+    /* 0x0F */ Delay,
+    /* 0x10 */ Upsample,
+    /* 0x11 */ DownMix6chTo2ch,
+    /* 0x12 */ Aux,
+    /* 0x13 */ DeviceSink,
+    /* 0x14 */ CircularBufferSink,
+    /* 0x15 */ Reverb,
+    /* 0x16 */ I3dl2Reverb,
+    /* 0x17 */ Performance,
+    /* 0x18 */ ClearMixBuffer,
+    /* 0x19 */ CopyMixBuffer,
+    /* 0x1A */ LightLimiterVersion1,
+    /* 0x1B */ LightLimiterVersion2,
+    /* 0x1C */ MultiTapBiquadFilter,
+    /* 0x1D */ Capture,
+};
+
+constexpr u32 CommandMagic{0xCAFEBABE};
+
+/**
+ * A command, generated by the host, and processed by the ADSP's AudioRenderer.
+ */
+struct ICommand {
+    virtual ~ICommand() = default;
+
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    virtual void Dump(const ADSP::CommandListProcessor& processor, std::string& string) = 0;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    virtual void Process(const ADSP::CommandListProcessor& processor) = 0;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    virtual bool Verify(const ADSP::CommandListProcessor& processor) = 0;
+
+    /// Command magic 0xCAFEBABE
+    u32 magic{};
+    /// Command enabled
+    bool enabled{};
+    /// Type of this command (see CommandId)
+    CommandId type{};
+    /// Size of this command
+    s16 size{};
+    /// Estimated processing time for this command
+    u32 estimated_process_time{};
+    /// Node id of the voice or mix this command was generated from
+    u32 node_id{};
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/mix/clear_mix.cpp

@@ -0,0 +1,24 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <string>
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/mix/clear_mix.h"
+
+namespace AudioCore::AudioRenderer {
+
+void ClearMixBufferCommand::Dump([[maybe_unused]] const ADSP::CommandListProcessor& processor,
+                                 std::string& string) {
+    string += fmt::format("ClearMixBufferCommand\n");
+}
+
+void ClearMixBufferCommand::Process(const ADSP::CommandListProcessor& processor) {
+    memset(processor.mix_buffers.data(), 0, processor.mix_buffers.size_bytes());
+}
+
+bool ClearMixBufferCommand::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/mix/clear_mix.h

@@ -0,0 +1,45 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <string>
+
+#include "audio_core/renderer/command/icommand.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command for a clearing the mix buffers.
+ * Used at the start of each command list.
+ */
+struct ClearMixBufferCommand : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+};
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/mix/copy_mix.cpp

@@ -0,0 +1,27 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/adsp/command_list_processor.h"
+#include "audio_core/renderer/command/mix/copy_mix.h"
+
+namespace AudioCore::AudioRenderer {
+
+void CopyMixBufferCommand::Dump([[maybe_unused]] const ADSP::CommandListProcessor& processor,
+                                std::string& string) {
+    string += fmt::format("CopyMixBufferCommand\n\tinput {:02X} output {:02X}\n", input_index,
+                          output_index);
+}
+
+void CopyMixBufferCommand::Process(const ADSP::CommandListProcessor& processor) {
+    auto output{processor.mix_buffers.subspan(output_index * processor.sample_count,
+                                              processor.sample_count)};
+    auto input{processor.mix_buffers.subspan(input_index * processor.sample_count,
+                                             processor.sample_count)};
+    std::memcpy(output.data(), input.data(), processor.sample_count * sizeof(s32));
+}
+
+bool CopyMixBufferCommand::Verify(const ADSP::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::AudioRenderer

src/audio_core/renderer/command/mix/copy_mix.h

@@ -0,0 +1,49 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <string>
+
+#include "audio_core/renderer/command/icommand.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+namespace ADSP {
+class CommandListProcessor;
+}
+
+/**
+ * AudioRenderer command for a copying a mix buffer from input to output.
+ */
+struct CopyMixBufferCommand : ICommand {
+    /**
+     * Print this command's information to a string.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @param string    - The string to print into.
+     */
+    void Dump(const ADSP::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const ADSP::CommandListProcessor& processor) override;
+
+    /**
+     * Verify this command's data is valid.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     * @return True if the command is valid, otherwise false.
+     */
+    bool Verify(const ADSP::CommandListProcessor& processor) override;
+
+    /// Input mix buffer index
+    s16 input_index;
+    /// Output mix buffer index
+    s16 output_index;
+};
+
+} // namespace AudioCore::AudioRenderer