Merge branch 'sdk20-rev15-audio-changes' into 'main'

feat: REV15 audio renderer + HID fix (v0.8.0)

See merge request citron/emulator!97

src/audio_core/CMakeLists.txt

@@ -30,15 +30,27 @@ add_library(audio_core STATIC
     audio_out_manager.h
     audio_manager.cpp
     audio_manager.h
+    audio_system_manager.cpp
+    audio_system_manager.h
+    audio_snoop_manager.cpp
+    audio_snoop_manager.h
+    common/audio_helpers.h
     common/audio_renderer_parameter.h
     common/common.h
     common/feature_support.h
+    common/fft.cpp
+    common/fft.h
+    common/loudness_calculator.cpp
+    common/loudness_calculator.h
     common/wave_buffer.h
     common/workbuffer_allocator.h
     device/audio_buffer.h
+    device/audio_buffer_list.h
     device/audio_buffers.h
     device/device_session.cpp
     device/device_session.h
+    device/shared_ring_buffer.cpp
+    device/shared_ring_buffer.h
     in/audio_in.cpp
     in/audio_in.h
     in/audio_in_system.cpp
@@ -85,6 +97,8 @@ add_library(audio_core STATIC
     renderer/command/effect/i3dl2_reverb.h
     renderer/command/effect/light_limiter.cpp
     renderer/command/effect/light_limiter.h
+    renderer/command/effect/limiter.cpp
+    renderer/command/effect/limiter.h
     renderer/command/effect/multi_tap_biquad_filter.cpp
     renderer/command/effect/multi_tap_biquad_filter.h
     renderer/command/effect/reverb.cpp
@@ -149,8 +163,13 @@ add_library(audio_core STATIC
     renderer/effect/i3dl2.h
     renderer/effect/light_limiter.cpp
     renderer/effect/light_limiter.h
+    renderer/effect/limiter.cpp
+    renderer/effect/limiter.h
     renderer/effect/reverb.h
     renderer/effect/reverb.cpp
+    renderer/final_output_recorder/final_output_recorder_buffer.h
+    renderer/final_output_recorder/final_output_recorder_system.cpp
+    renderer/final_output_recorder/final_output_recorder_system.h
     renderer/mix/mix_context.cpp
     renderer/mix/mix_context.h
     renderer/mix/mix_info.cpp
@@ -218,6 +237,7 @@ if (MSVC)
         /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
         /we4800 # Implicit conversion from 'type' to bool. Possible information loss
+        /wd2375 # Disable C2375: '__builtin_assume_aligned': redefinition (MSVC 14.44+ issue)
     )
 else()
     target_compile_options(audio_core PRIVATE

src/audio_core/audio_snoop_manager.cpp

@@ -0,0 +1,78 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_snoop_manager.h"
+#include "common/logging/log.h"
+#include "core/core.h"
+#include "core/hle/result.h"
+
+namespace AudioCore {
+
+AudioSnoopManager::AudioSnoopManager(Core::System& system_) : system{system_} {}
+AudioSnoopManager::~AudioSnoopManager() = default;
+
+Result AudioSnoopManager::GetDspStatistics(DspStatistics& statistics_out) {
+    std::scoped_lock lock{mutex};
+
+    if (!statistics_enabled) {
+        LOG_DEBUG(Service_Audio, "DSP statistics not enabled");
+        statistics_out = {};
+        return ResultSuccess;
+    }
+
+    statistics_out = statistics;
+    return ResultSuccess;
+}
+
+Result AudioSnoopManager::SetDspStatisticsParameter(bool enabled) {
+    std::scoped_lock lock{mutex};
+
+    LOG_DEBUG(Service_Audio, "Set DSP statistics enabled: {}", enabled);
+    statistics_enabled = enabled;
+
+    if (!enabled) {
+        // Clear statistics when disabled
+        statistics = {};
+    }
+
+    return ResultSuccess;
+}
+
+Result AudioSnoopManager::GetDspStatisticsParameter(bool& enabled) {
+    std::scoped_lock lock{mutex};
+    enabled = statistics_enabled;
+    return ResultSuccess;
+}
+
+Result AudioSnoopManager::GetAppletStateSummaries(std::span<AppletStateSummary> summaries,
+                                                   u32& count) {
+    std::scoped_lock lock{mutex};
+
+    // For now, return empty summaries
+    // A full implementation would track active audio sessions per applet
+    count = 0;
+    return ResultSuccess;
+}
+
+void AudioSnoopManager::UpdateStatistics(u64 cycles_elapsed, u32 active_voices,
+                                          u32 dropped_commands) {
+    if (!statistics_enabled) {
+        return;
+    }
+
+    std::scoped_lock lock{mutex};
+
+    statistics.total_cycles += cycles_elapsed;
+    statistics.active_cycles += cycles_elapsed;
+    statistics.command_drop_count += dropped_commands;
+
+    // Calculate usage percentages (simplified)
+    if (statistics.total_cycles > 0) {
+        statistics.cpu_usage_percent =
+            static_cast<f32>(statistics.active_cycles) / static_cast<f32>(statistics.total_cycles) *
+            100.0f;
+        statistics.dsp_usage_percent = statistics.cpu_usage_percent * 0.5f; // Estimate
+    }
+}
+
+} // namespace AudioCore

src/audio_core/audio_snoop_manager.h

@@ -0,0 +1,100 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <mutex>
+#include <span>
+
+#include "common/common_types.h"
+#include "core/hle/result.h"
+
+namespace Core {
+class System;
+}
+
+namespace AudioCore {
+
+/**
+ * Manages DSP statistics and performance monitoring.
+ */
+class AudioSnoopManager {
+public:
+    struct DspStatistics {
+        /* 0x00 */ u64 total_cycles;
+        /* 0x08 */ u64 active_cycles;
+        /* 0x10 */ u32 voice_drop_count;
+        /* 0x14 */ u32 command_drop_count;
+        /* 0x18 */ u32 buffer_underrun_count;
+        /* 0x1C */ u32 buffer_overrun_count;
+        /* 0x20 */ f32 cpu_usage_percent;
+        /* 0x24 */ f32 dsp_usage_percent;
+    };
+    static_assert(sizeof(DspStatistics) == 0x28, "DspStatistics has the wrong size!");
+
+    struct AppletStateSummary {
+        /* 0x00 */ u64 applet_resource_user_id;
+        /* 0x08 */ u32 audio_in_active_count;
+        /* 0x0C */ u32 audio_out_active_count;
+        /* 0x10 */ u32 audio_renderer_active_count;
+        /* 0x14 */ u32 final_output_recorder_active_count;
+        /* 0x18 */ u32 total_active_count;
+        /* 0x1C */ u32 reserved;
+    };
+    static_assert(sizeof(AppletStateSummary) == 0x20, "AppletStateSummary has the wrong size!");
+
+    explicit AudioSnoopManager(Core::System& system);
+    ~AudioSnoopManager();
+
+    /**
+     * Get DSP statistics.
+     *
+     * @param statistics - Output statistics structure.
+     * @return Result code.
+     */
+    Result GetDspStatistics(DspStatistics& statistics);
+
+    /**
+     * Set DSP statistics parameter.
+     *
+     * @param enabled - Enable or disable statistics gathering.
+     * @return Result code.
+     */
+    Result SetDspStatisticsParameter(bool enabled);
+
+    /**
+     * Get DSP statistics parameter.
+     *
+     * @param enabled - Output enabled state.
+     * @return Result code.
+     */
+    Result GetDspStatisticsParameter(bool& enabled);
+
+    /**
+     * Get applet state summaries.
+     *
+     * @param summaries - Output array of summaries.
+     * @param count - Output count of summaries.
+     * @return Result code.
+     */
+    Result GetAppletStateSummaries(std::span<AppletStateSummary> summaries, u32& count);
+
+    /**
+     * Update statistics (called periodically by the audio system).
+     *
+     * @param cycles_elapsed - Number of CPU cycles elapsed.
+     * @param active_voices - Number of active voices.
+     * @param dropped_commands - Number of dropped commands.
+     */
+    void UpdateStatistics(u64 cycles_elapsed, u32 active_voices, u32 dropped_commands);
+
+private:
+    Core::System& system;
+    std::mutex mutex;
+
+    DspStatistics statistics{};
+    bool statistics_enabled{false};
+};
+
+} // namespace AudioCore

src/audio_core/audio_system_manager.cpp

@@ -0,0 +1,128 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <algorithm>
+
+#include "audio_core/audio_system_manager.h"
+#include "common/logging/log.h"
+#include "core/core.h"
+#include "core/hle/result.h"
+#include "core/hle/service/audio/errors.h"
+
+namespace AudioCore {
+
+AudioSystemManager::AudioSystemManager(Core::System& system_) : system{system_} {}
+AudioSystemManager::~AudioSystemManager() = default;
+
+Result AudioSystemManager::RegisterAppletResourceUserId(u64 applet_resource_user_id) {
+    std::scoped_lock lock{mutex};
+
+    if (registered_count >= MaxAppletResourceUserIds) {
+        LOG_ERROR(Service_Audio, "Maximum applet resource user IDs registered");
+        return Service::Audio::ResultInvalidHandle;
+    }
+
+    // Check if already registered
+    for (size_t i = 0; i < registered_count; i++) {
+        if (registered_ids[i] == applet_resource_user_id) {
+            return ResultSuccess; // Already registered
+        }
+    }
+
+    registered_ids[registered_count++] = applet_resource_user_id;
+    LOG_DEBUG(Service_Audio, "Registered applet resource user ID: {}", applet_resource_user_id);
+    return ResultSuccess;
+}
+
+Result AudioSystemManager::UnregisterAppletResourceUserId(u64 applet_resource_user_id) {
+    std::scoped_lock lock{mutex};
+
+    for (size_t i = 0; i < registered_count; i++) {
+        if (registered_ids[i] == applet_resource_user_id) {
+            // Remove by shifting remaining elements
+            for (size_t j = i; j < registered_count - 1; j++) {
+                registered_ids[j] = registered_ids[j + 1];
+            }
+            registered_count--;
+            LOG_DEBUG(Service_Audio, "Unregistered applet resource user ID: {}",
+                      applet_resource_user_id);
+            return ResultSuccess;
+        }
+    }
+
+    LOG_WARNING(Service_Audio, "Applet resource user ID not found: {}", applet_resource_user_id);
+    return Service::Audio::ResultInvalidHandle;
+}
+
+Result AudioSystemManager::RequestSuspendAudio(u64 applet_resource_user_id) {
+    std::scoped_lock lock{mutex};
+
+    LOG_DEBUG(Service_Audio, "Suspending audio for applet resource user ID: {}",
+              applet_resource_user_id);
+    audio_suspended = true;
+    return ResultSuccess;
+}
+
+Result AudioSystemManager::RequestResumeAudio(u64 applet_resource_user_id) {
+    std::scoped_lock lock{mutex};
+
+    LOG_DEBUG(Service_Audio, "Resuming audio for applet resource user ID: {}",
+              applet_resource_user_id);
+    audio_suspended = false;
+    return ResultSuccess;
+}
+
+Result AudioSystemManager::GetAudioInputProcessMasterVolume(f32& volume) {
+    std::scoped_lock lock{mutex};
+    volume = input_master_volume;
+    return ResultSuccess;
+}
+
+Result AudioSystemManager::SetAudioInputProcessMasterVolume(f32 volume) {
+    std::scoped_lock lock{mutex};
+    input_master_volume = std::clamp(volume, 0.0f, 1.0f);
+    LOG_DEBUG(Service_Audio, "Set audio input master volume: {}", input_master_volume);
+    return ResultSuccess;
+}
+
+Result AudioSystemManager::GetAudioOutputProcessMasterVolume(f32& volume) {
+    std::scoped_lock lock{mutex};
+    volume = output_master_volume;
+    return ResultSuccess;
+}
+
+Result AudioSystemManager::SetAudioOutputProcessMasterVolume(f32 volume) {
+    std::scoped_lock lock{mutex};
+    output_master_volume = std::clamp(volume, 0.0f, 1.0f);
+    LOG_DEBUG(Service_Audio, "Set audio output master volume: {}", output_master_volume);
+    return ResultSuccess;
+}
+
+Result AudioSystemManager::GetAudioOutputProcessRecordVolume(f32& volume) {
+    std::scoped_lock lock{mutex};
+    volume = output_record_volume;
+    return ResultSuccess;
+}
+
+Result AudioSystemManager::SetAudioOutputProcessRecordVolume(f32 volume) {
+    std::scoped_lock lock{mutex};
+    output_record_volume = std::clamp(volume, 0.0f, 1.0f);
+    LOG_DEBUG(Service_Audio, "Set audio output record volume: {}", output_record_volume);
+    return ResultSuccess;
+}
+
+Result AudioSystemManager::RequestSuspendAudioForDebug() {
+    std::scoped_lock lock{mutex};
+    LOG_DEBUG(Service_Audio, "Suspending audio for debug");
+    debug_suspended = true;
+    return ResultSuccess;
+}
+
+Result AudioSystemManager::RequestResumeAudioForDebug() {
+    std::scoped_lock lock{mutex};
+    LOG_DEBUG(Service_Audio, "Resuming audio for debug");
+    debug_suspended = false;
+    return ResultSuccess;
+}
+
+} // namespace AudioCore

src/audio_core/audio_system_manager.h

@@ -0,0 +1,137 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <mutex>
+
+#include "common/common_types.h"
+#include "core/hle/result.h"
+
+namespace Core {
+class System;
+}
+
+namespace AudioCore {
+
+/**
+ * Manages audio system state for applets including suspend/resume and volume control.
+ */
+class AudioSystemManager {
+public:
+    explicit AudioSystemManager(Core::System& system);
+    ~AudioSystemManager();
+
+    /**
+     * Register an applet resource user ID.
+     *
+     * @param applet_resource_user_id - Applet resource user ID to register.
+     * @return Result code.
+     */
+    Result RegisterAppletResourceUserId(u64 applet_resource_user_id);
+
+    /**
+     * Unregister an applet resource user ID.
+     *
+     * @param applet_resource_user_id - Applet resource user ID to unregister.
+     * @return Result code.
+     */
+    Result UnregisterAppletResourceUserId(u64 applet_resource_user_id);
+
+    /**
+     * Request audio suspension.
+     *
+     * @param applet_resource_user_id - Applet resource user ID.
+     * @return Result code.
+     */
+    Result RequestSuspendAudio(u64 applet_resource_user_id);
+
+    /**
+     * Request audio resumption.
+     *
+     * @param applet_resource_user_id - Applet resource user ID.
+     * @return Result code.
+     */
+    Result RequestResumeAudio(u64 applet_resource_user_id);
+
+    /**
+     * Get audio input process master volume.
+     *
+     * @param volume - Output volume value.
+     * @return Result code.
+     */
+    Result GetAudioInputProcessMasterVolume(f32& volume);
+
+    /**
+     * Set audio input process master volume.
+     *
+     * @param volume - Volume value to set.
+     * @return Result code.
+     */
+    Result SetAudioInputProcessMasterVolume(f32 volume);
+
+    /**
+     * Get audio output process master volume.
+     *
+     * @param volume - Output volume value.
+     * @return Result code.
+     */
+    Result GetAudioOutputProcessMasterVolume(f32& volume);
+
+    /**
+     * Set audio output process master volume.
+     *
+     * @param volume - Volume value to set.
+     * @return Result code.
+     */
+    Result SetAudioOutputProcessMasterVolume(f32 volume);
+
+    /**
+     * Get audio output process record volume.
+     *
+     * @param volume - Output volume value.
+     * @return Result code.
+     */
+    Result GetAudioOutputProcessRecordVolume(f32& volume);
+
+    /**
+     * Set audio output process record volume.
+     *
+     * @param volume - Volume value to set.
+     * @return Result code.
+     */
+    Result SetAudioOutputProcessRecordVolume(f32 volume);
+
+    /**
+     * Request audio suspension for debugging.
+     *
+     * @return Result code.
+     */
+    Result RequestSuspendAudioForDebug();
+
+    /**
+     * Request audio resumption for debugging.
+     *
+     * @return Result code.
+     */
+    Result RequestResumeAudioForDebug();
+
+private:
+    static constexpr size_t MaxAppletResourceUserIds = 8;
+
+    Core::System& system;
+    std::mutex mutex;
+
+    std::array<u64, MaxAppletResourceUserIds> registered_ids{};
+    size_t registered_count{0};
+
+    f32 input_master_volume{1.0f};
+    f32 output_master_volume{1.0f};
+    f32 output_record_volume{1.0f};
+
+    bool audio_suspended{false};
+    bool debug_suspended{false};
+};
+
+} // namespace AudioCore

src/audio_core/common/audio_helpers.h

@@ -0,0 +1,165 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron 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"
+
+namespace AudioCore {
+
+/**
+ * ADPCM context structure
+ */
+struct AdpcmContext {
+    u16 header;
+    s16 yn0;
+    s16 yn1;
+};
+
+/**
+ * Parse ADPCM header information.
+ *
+ * @param context - Output context to receive parsed information.
+ * @param data    - Input ADPCM data.
+ */
+inline void ParseAdpcmHeader(AdpcmContext& context, std::span<const u8> data) {
+    if (data.size() < 4) {
+        context = {};
+        return;
+    }
+
+    context.header = static_cast<u16>((data[0] << 8) | data[1]);
+    context.yn0 = static_cast<s16>((data[2] << 8) | data[3]);
+    context.yn1 = 0;
+}
+
+/**
+ * Get the byte size for a sample format.
+ *
+ * @param format - Sample format.
+ * @return Size in bytes.
+ */
+constexpr size_t GetSampleByteSize(SampleFormat format) {
+    return GetSampleFormatByteSize(format);
+}
+
+/**
+ * Initialize audio in parameter with defaults.
+ *
+ * @param params - Parameter structure to initialize.
+ */
+inline void InitializeAudioInParameter(auto& params) {
+    params.sample_rate = TargetSampleRate;
+    params.channel_count = 2;
+    params.reserved = 0;
+}
+
+/**
+ * Initialize audio out parameter with defaults.
+ *
+ * @param params - Parameter structure to initialize.
+ */
+inline void InitializeAudioOutParameter(auto& params) {
+    params.sample_rate = TargetSampleRate;
+    params.channel_count = 2;
+    params.reserved = 0;
+}
+
+/**
+ * Initialize final output recorder parameter with defaults.
+ *
+ * @param params - Parameter structure to initialize.
+ */
+inline void InitializeFinalOutputRecorderParameter(auto& params) {
+    params.sample_rate = TargetSampleRate;
+    params.channel_count = 2;
+    params.reserved = 0;
+}
+
+/**
+ * Set audio in buffer information.
+ *
+ * @param buffer       - Buffer to set information for.
+ * @param data_address - Address of the sample data.
+ * @param size         - Size of the buffer in bytes.
+ */
+inline void SetAudioInBufferInfo(auto& buffer, VAddr data_address, u64 size) {
+    buffer.samples = data_address;
+    buffer.capacity = size;
+    buffer.size = 0;
+    buffer.offset = 0;
+}
+
+/**
+ * Set audio out buffer information.
+ *
+ * @param buffer       - Buffer to set information for.
+ * @param data_address - Address of the sample data.
+ * @param size         - Size of the buffer in bytes.
+ */
+inline void SetAudioOutBufferInfo(auto& buffer, VAddr data_address, u64 size) {
+    buffer.samples = data_address;
+    buffer.capacity = size;
+    buffer.size = size;
+    buffer.offset = 0;
+}
+
+/**
+ * Get the data pointer from an audio buffer.
+ *
+ * @param buffer - Buffer to get data from.
+ * @return Address of the sample data.
+ */
+inline VAddr GetAudioBufferDataPointer(const auto& buffer) {
+    return buffer.samples;
+}
+
+/**
+ * Get the data size from an audio buffer.
+ *
+ * @param buffer - Buffer to get size from.
+ * @return Size of valid data in bytes.
+ */
+inline u64 GetAudioBufferDataSize(const auto& buffer) {
+    return buffer.size;
+}
+
+/**
+ * Get the buffer size from an audio buffer.
+ *
+ * @param buffer - Buffer to get size from.
+ * @return Total buffer capacity in bytes.
+ */
+inline u64 GetAudioBufferBufferSize(const auto& buffer) {
+    return buffer.capacity;
+}
+
+/**
+ * Get the required buffer size for final output recorder.
+ *
+ * @param sample_count - Number of samples.
+ * @param channel_count - Number of channels.
+ * @return Required buffer size in bytes.
+ */
+inline u64 GetFinalOutputRecorderWorkBufferSize(u32 sample_count, u32 channel_count) {
+    constexpr size_t buffer_header_size = 0x100;
+    const size_t sample_data_size = sample_count * channel_count * sizeof(s16);
+    const size_t buffer_count = 32;
+    return buffer_header_size + (sample_data_size * buffer_count);
+}
+
+/**
+ * Initialize final output recorder work buffer parameter.
+ *
+ * @param address - Work buffer address.
+ * @param size    - Work buffer size.
+ */
+inline void InitializeFinalOutputRecorderWorkBufferParameter(VAddr address, u64 size) {
+    // Work buffer is just a memory region, no initialization needed
+}
+
+} // namespace AudioCore

src/audio_core/common/feature_support.h

@@ -1,4 +1,5 @@
 // SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
@@ -13,7 +14,7 @@
 #include "common/polyfill_ranges.h"
 
 namespace AudioCore {
-constexpr u32 CurrentRevision = 13;
+constexpr u32 CurrentRevision = 15;
 
 enum class SupportTags {
     CommandProcessingTimeEstimatorVersion4,
@@ -46,6 +47,8 @@ enum class SupportTags {
     I3dl2ReverbChannelMappingChange,
     CompressorStatistics,
     SplitterPrevVolumeReset,
+    SplitterDestinationV2b,
+    VoiceInParameterV2,
 
     // Not a real tag, just here to get the count.
     Size
@@ -91,6 +94,9 @@ constexpr bool CheckFeatureSupported(SupportTags tag, u32 user_revision) {
             {SupportTags::I3dl2ReverbChannelMappingChange, 11},
             {SupportTags::CompressorStatistics, 13},
             {SupportTags::SplitterPrevVolumeReset, 13},
+            {SupportTags::DeviceApiVersion2, 13},
+            {SupportTags::SplitterDestinationV2b, 15},
+            {SupportTags::VoiceInParameterV2, 15},
         }};
 
     const auto& feature =

src/audio_core/common/fft.cpp

@@ -0,0 +1,167 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <algorithm>
+#include <cmath>
+#include <numbers>
+
+#include "audio_core/common/fft.h"
+#include "common/logging/log.h"
+
+namespace AudioCore {
+
+FFT::~FFT() = default;
+
+size_t FFT::GetWorkBufferSize(u32 sample_count, ProcessMode mode) {
+    // Work buffer needs space for complex samples and temporary data
+    const size_t complex_size = sample_count * sizeof(std::complex<f32>);
+    const size_t temp_size = sample_count * sizeof(std::complex<f32>);
+    return complex_size + temp_size;
+}
+
+size_t FFT::GetWorkBufferAlignment() {
+    return 64; // Cache line alignment
+}
+
+bool FFT::Initialize(u32 sample_count, ProcessMode mode, void* work_buffer,
+                     size_t work_buffer_size) {
+    // Verify sample count is power of 2
+    if ((sample_count & (sample_count - 1)) != 0) {
+        LOG_ERROR(Audio, "FFT sample count must be power of 2, got {}", sample_count);
+        return false;
+    }
+
+    if (work_buffer == nullptr || work_buffer_size < GetWorkBufferSize(sample_count, mode)) {
+        LOG_ERROR(Audio, "FFT work buffer too small");
+        return false;
+    }
+
+    sample_count_ = sample_count;
+    mode_ = mode;
+    initialized_ = true;
+    return true;
+}
+
+void FFT::ProcessRealToComplex(std::span<std::complex<f32>> output, std::span<const f32> input,
+                                u32 sample_count) {
+    if (!initialized_ || sample_count != sample_count_) {
+        LOG_ERROR(Audio, "FFT not initialized or sample count mismatch");
+        return;
+    }
+
+    // Convert real input to complex
+    std::vector<std::complex<f32>> temp(sample_count);
+    for (u32 i = 0; i < sample_count; i++) {
+        temp[i] = std::complex<f32>(input[i], 0.0f);
+    }
+
+    // Perform FFT
+    FFTInternal(temp, sample_count, false);
+
+    // Copy to output (only first half + 1 due to symmetry)
+    const u32 output_count = sample_count / 2 + 1;
+    std::copy_n(temp.begin(), output_count, output.begin());
+}
+
+void FFT::ProcessComplexToReal(std::span<f32> output, std::span<const std::complex<f32>> input,
+                                u32 sample_count) {
+    if (!initialized_ || sample_count != sample_count_) {
+        LOG_ERROR(Audio, "FFT not initialized or sample count mismatch");
+        return;
+    }
+
+    // Reconstruct full complex spectrum (conjugate symmetry)
+    std::vector<std::complex<f32>> temp(sample_count);
+    const u32 half = sample_count / 2;
+
+    // Copy provided values
+    for (u32 i = 0; i <= half; i++) {
+        temp[i] = input[i];
+    }
+
+    // Mirror with conjugate
+    for (u32 i = 1; i < half; i++) {
+        temp[sample_count - i] = std::conj(temp[i]);
+    }
+
+    // Perform inverse FFT
+    FFTInternal(temp, sample_count, true);
+
+    // Extract real part
+    for (u32 i = 0; i < sample_count; i++) {
+        output[i] = temp[i].real();
+    }
+}
+
+void FFT::ProcessComplexToComplex(std::span<std::complex<f32>> output,
+                                   std::span<const std::complex<f32>> input, u32 sample_count,
+                                   bool inverse) {
+    if (!initialized_ || sample_count != sample_count_) {
+        LOG_ERROR(Audio, "FFT not initialized or sample count mismatch");
+        return;
+    }
+
+    // Copy input to output
+    std::copy_n(input.begin(), sample_count, output.begin());
+
+    // Perform FFT
+    FFTInternal(output, sample_count, inverse);
+}
+
+void FFT::BitReverseCopy(std::span<std::complex<f32>> output,
+                          std::span<const std::complex<f32>> input, u32 size) {
+    const u32 bits = std::bit_width(size - 1);
+
+    for (u32 i = 0; i < size; i++) {
+        u32 j = 0;
+        for (u32 b = 0; b < bits; b++) {
+            if (i & (1u << b)) {
+                j |= 1u << (bits - 1 - b);
+            }
+        }
+        output[j] = input[i];
+    }
+}
+
+void FFT::FFTInternal(std::span<std::complex<f32>> data, u32 size, bool inverse) {
+    // Cooley-Tukey FFT algorithm
+    std::vector<std::complex<f32>> temp(size);
+    BitReverseCopy(temp, data, size);
+    std::copy(temp.begin(), temp.end(), data.begin());
+
+    const f32 direction = inverse ? 1.0f : -1.0f;
+    const f32 scale = inverse ? (1.0f / size) : 1.0f;
+
+    // FFT stages
+    const u32 log2_size = static_cast<u32>(std::bit_width(size - 1u));
+    for (u32 s = 1; s <= log2_size; s++) {
+        const u32 m = 1u << s;
+        const u32 m2 = m / 2;
+
+        const std::complex<f32> wm = std::exp(std::complex<f32>(
+            0.0f, direction * 2.0f * std::numbers::pi_v<f32> / m));
+
+        for (u32 k = 0; k < size; k += m) {
+            std::complex<f32> w = 1.0f;
+
+            for (u32 j = 0; j < m2; j++) {
+                const std::complex<f32> t = w * data[k + j + m2];
+                const std::complex<f32> u = data[k + j];
+
+                data[k + j] = u + t;
+                data[k + j + m2] = u - t;
+
+                w *= wm;
+            }
+        }
+    }
+
+    // Apply scaling for inverse transform
+    if (inverse) {
+        for (u32 i = 0; i < size; i++) {
+            data[i] *= scale;
+        }
+    }
+}
+
+} // namespace AudioCore

src/audio_core/common/fft.h

@@ -0,0 +1,99 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <complex>
+#include <memory>
+#include <span>
+
+#include "common/common_types.h"
+
+namespace AudioCore {
+
+/**
+ * FFT implementation for audio processing.
+ * Simplified version compatible with Nintendo SDK FFT functions.
+ */
+class FFT {
+public:
+    enum class ProcessMode {
+        RealToComplex,
+        ComplexToReal,
+        ComplexToComplex,
+    };
+
+    FFT() = default;
+    ~FFT();
+
+    /**
+     * Get the required work buffer size for FFT processing.
+     *
+     * @param sample_count - Number of samples to process (must be power of 2).
+     * @param mode - Processing mode.
+     * @return Required buffer size in bytes.
+     */
+    static size_t GetWorkBufferSize(u32 sample_count, ProcessMode mode);
+
+    /**
+     * Get the required work buffer alignment.
+     *
+     * @return Required alignment in bytes.
+     */
+    static size_t GetWorkBufferAlignment();
+
+    /**
+     * Initialize the FFT processor.
+     *
+     * @param sample_count - Number of samples to process (must be power of 2).
+     * @param mode - Processing mode.
+     * @param work_buffer - Work buffer for temporary data.
+     * @param work_buffer_size - Size of work buffer.
+     * @return True if initialization succeeded.
+     */
+    bool Initialize(u32 sample_count, ProcessMode mode, void* work_buffer,
+                    size_t work_buffer_size);
+
+    /**
+     * Process real-to-complex FFT (forward transform).
+     *
+     * @param output - Output complex samples.
+     * @param input - Input real samples.
+     * @param sample_count - Number of samples.
+     */
+    void ProcessRealToComplex(std::span<std::complex<f32>> output, std::span<const f32> input,
+                               u32 sample_count);
+
+    /**
+     * Process complex-to-real FFT (inverse transform).
+     *
+     * @param output - Output real samples.
+     * @param input - Input complex samples.
+     * @param sample_count - Number of samples.
+     */
+    void ProcessComplexToReal(std::span<f32> output, std::span<const std::complex<f32>> input,
+                               u32 sample_count);
+
+    /**
+     * Process complex-to-complex FFT.
+     *
+     * @param output - Output complex samples.
+     * @param input - Input complex samples.
+     * @param sample_count - Number of samples.
+     * @param inverse - True for inverse transform, false for forward.
+     */
+    void ProcessComplexToComplex(std::span<std::complex<f32>> output,
+                                  std::span<const std::complex<f32>> input, u32 sample_count,
+                                  bool inverse);
+
+private:
+    void BitReverseCopy(std::span<std::complex<f32>> output,
+                        std::span<const std::complex<f32>> input, u32 size);
+    void FFTInternal(std::span<std::complex<f32>> data, u32 size, bool inverse);
+
+    u32 sample_count_{0};
+    ProcessMode mode_{ProcessMode::RealToComplex};
+    bool initialized_{false};
+};
+
+} // namespace AudioCore

src/audio_core/common/loudness_calculator.cpp

@@ -0,0 +1,208 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <algorithm>
+#include <cmath>
+#include <numbers>
+
+#include "audio_core/common/loudness_calculator.h"
+#include "common/logging/log.h"
+
+namespace AudioCore {
+
+LoudnessCalculator::LoudnessCalculator() = default;
+LoudnessCalculator::~LoudnessCalculator() {
+    Finalize();
+}
+
+bool LoudnessCalculator::Initialize(const Parameters& params) {
+    if (params.channel_count == 0 || params.channel_count > MaxChannels) {
+        LOG_ERROR(Audio, "Invalid channel count: {}", params.channel_count);
+        return false;
+    }
+
+    if (params.sample_rate == 0) {
+        LOG_ERROR(Audio, "Invalid sample rate: {}", params.sample_rate);
+        return false;
+    }
+
+    params_ = params;
+    initialized_ = true;
+
+    InitializeKWeightingFilter();
+    Reset();
+
+    return true;
+}
+
+void LoudnessCalculator::Finalize() {
+    initialized_ = false;
+}
+
+void LoudnessCalculator::Reset() {
+    if (!initialized_) {
+        return;
+    }
+
+    momentary_loudness_ = -70.0f;
+    short_term_loudness_ = -70.0f;
+    integrated_loudness_ = -70.0f;
+    loudness_range_ = 0.0f;
+    integrated_sum_ = 0.0f;
+    integrated_count_ = 0;
+    buffer_index_ = 0;
+
+    momentary_buffer_.fill(0.0f);
+    short_term_buffer_.fill(0.0f);
+
+    // Reset filter state
+    std::fill(std::begin(k_filter_.z1_shelf), std::end(k_filter_.z1_shelf), 0.0f);
+    std::fill(std::begin(k_filter_.z2_shelf), std::end(k_filter_.z2_shelf), 0.0f);
+    std::fill(std::begin(k_filter_.z1_hp), std::end(k_filter_.z1_hp), 0.0f);
+    std::fill(std::begin(k_filter_.z2_hp), std::end(k_filter_.z2_hp), 0.0f);
+}
+
+void LoudnessCalculator::InitializeKWeightingFilter() {
+    // K-weighting filter coefficients for 48kHz
+    // Shelf filter (high-shelf +4dB at high frequencies)
+    const f32 f0_shelf = 1681.974450955533f;
+    const f32 Q_shelf = 0.7071752369554193f;
+    const f32 K_shelf = std::tan(std::numbers::pi_v<f32> * f0_shelf / params_.sample_rate);
+    const f32 Vh_shelf = std::pow(10.0f, 4.0f / 20.0f);
+    const f32 Vb_shelf = std::pow(Vh_shelf, 0.4996667741545416f);
+
+    const f32 a0_shelf = 1.0f + K_shelf / Q_shelf + K_shelf * K_shelf;
+    k_filter_.b0_shelf = (Vh_shelf + Vb_shelf * K_shelf / Q_shelf + K_shelf * K_shelf) / a0_shelf;
+    k_filter_.b1_shelf = 2.0f * (K_shelf * K_shelf - Vh_shelf) / a0_shelf;
+    k_filter_.b2_shelf = (Vh_shelf - Vb_shelf * K_shelf / Q_shelf + K_shelf * K_shelf) / a0_shelf;
+    k_filter_.a1_shelf = 2.0f * (K_shelf * K_shelf - 1.0f) / a0_shelf;
+    k_filter_.a2_shelf = (1.0f - K_shelf / Q_shelf + K_shelf * K_shelf) / a0_shelf;
+
+    // High-pass filter (48Hz cutoff)
+    const f32 f0_hp = 38.13547087602444f;
+    const f32 Q_hp = 0.5003270373238773f;
+    const f32 K_hp = std::tan(std::numbers::pi_v<f32> * f0_hp / params_.sample_rate);
+
+    const f32 a0_hp = 1.0f + K_hp / Q_hp + K_hp * K_hp;
+    k_filter_.b0_hp = 1.0f / a0_hp;
+    k_filter_.b1_hp = -2.0f / a0_hp;
+    k_filter_.b2_hp = 1.0f / a0_hp;
+    k_filter_.a1_hp = 2.0f * (K_hp * K_hp - 1.0f) / a0_hp;
+    k_filter_.a2_hp = (1.0f - K_hp / Q_hp + K_hp * K_hp) / a0_hp;
+}
+
+f32 LoudnessCalculator::ApplyKWeighting(f32 sample, u32 channel) {
+    // Apply shelf filter
+    const f32 out_shelf = k_filter_.b0_shelf * sample + k_filter_.b1_shelf * k_filter_.z1_shelf[channel] +
+                          k_filter_.b2_shelf * k_filter_.z2_shelf[channel] -
+                          k_filter_.a1_shelf * k_filter_.z1_shelf[channel] -
+                          k_filter_.a2_shelf * k_filter_.z2_shelf[channel];
+
+    k_filter_.z2_shelf[channel] = k_filter_.z1_shelf[channel];
+    k_filter_.z1_shelf[channel] = sample;
+
+    // Apply high-pass filter
+    const f32 out_hp = k_filter_.b0_hp * out_shelf + k_filter_.b1_hp * k_filter_.z1_hp[channel] +
+                       k_filter_.b2_hp * k_filter_.z2_hp[channel] -
+                       k_filter_.a1_hp * k_filter_.z1_hp[channel] -
+                       k_filter_.a2_hp * k_filter_.z2_hp[channel];
+
+    k_filter_.z2_hp[channel] = k_filter_.z1_hp[channel];
+    k_filter_.z1_hp[channel] = out_shelf;
+
+    return out_hp;
+}
+
+void LoudnessCalculator::Analyze(std::span<const f32> samples, u32 sample_count) {
+    if (!initialized_) {
+        return;
+    }
+
+    const u32 total_samples = sample_count * params_.channel_count;
+    if (samples.size() < total_samples) {
+        return;
+    }
+
+    // Process samples
+    for (u32 i = 0; i < sample_count; i++) {
+        f32 sum_square = 0.0f;
+
+        // Apply K-weighting and calculate mean square for each channel
+        for (u32 ch = 0; ch < params_.channel_count; ch++) {
+            const f32 sample = samples[i * params_.channel_count + ch];
+            const f32 weighted = ApplyKWeighting(sample, ch);
+
+            // Channel weighting (LFE channel has less weight)
+            const f32 weight = (ch == 3) ? 0.0f : 1.0f; // Channel 3 is typically LFE
+            sum_square += weight * weighted * weighted;
+        }
+
+        // Calculate mean square
+        const f32 mean_square = sum_square / params_.channel_count;
+
+        // Update buffers
+        momentary_buffer_[buffer_index_ % momentary_buffer_.size()] = mean_square;
+        short_term_buffer_[buffer_index_ % short_term_buffer_.size()] = mean_square;
+        buffer_index_++;
+
+        // Update integrated measurement
+        integrated_sum_ += mean_square;
+        integrated_count_++;
+    }
+
+    // Calculate momentary loudness (last 400ms)
+    const size_t momentary_samples = std::min(buffer_index_,
+        static_cast<size_t>(params_.sample_rate * 0.4f));
+    f32 momentary_sum = 0.0f;
+    for (size_t i = 0; i < momentary_samples; i++) {
+        momentary_sum += momentary_buffer_[i];
+    }
+    momentary_loudness_ = CalculateLoudness(momentary_sum / momentary_samples);
+
+    // Calculate short-term loudness (last 3s)
+    const size_t short_term_samples = std::min(buffer_index_,
+        static_cast<size_t>(params_.sample_rate * 3.0f));
+    f32 short_term_sum = 0.0f;
+    for (size_t i = 0; i < short_term_samples; i++) {
+        short_term_sum += short_term_buffer_[i];
+    }
+    short_term_loudness_ = CalculateLoudness(short_term_sum / short_term_samples);
+
+    // Calculate integrated loudness
+    if (integrated_count_ > 0) {
+        integrated_loudness_ = CalculateLoudness(integrated_sum_ / integrated_count_);
+    }
+}
+
+f32 LoudnessCalculator::CalculateLoudness(f32 mean_square) {
+    if (mean_square <= 0.0f) {
+        return -70.0f; // Silence
+    }
+
+    // Convert to LUFS: -0.691 + 10*log10(mean_square)
+    return -0.691f + 10.0f * std::log10(mean_square);
+}
+
+f32 LoudnessCalculator::GetMomentaryLoudness() const {
+    return momentary_loudness_;
+}
+
+f32 LoudnessCalculator::GetShortTermLoudness() const {
+    return short_term_loudness_;
+}
+
+f32 LoudnessCalculator::GetIntegratedLoudness() const {
+    return integrated_loudness_;
+}
+
+f32 LoudnessCalculator::GetLoudnessRange() const {
+    return loudness_range_;
+}
+
+void LoudnessCalculator::SetMomentaryLoudnessLpfTc(f32 time_constant) {
+    if (initialized_) {
+        params_.momentary_time_constant = time_constant;
+    }
+}
+
+} // namespace AudioCore

src/audio_core/common/loudness_calculator.h

@@ -0,0 +1,131 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <memory>
+#include <span>
+
+#include "common/common_types.h"
+
+namespace AudioCore {
+
+/**
+ * Loudness calculator following ITU-R BS.1770 standard.
+ * Used for measuring audio loudness in LUFS (Loudness Units Full Scale).
+ */
+class LoudnessCalculator {
+public:
+    static constexpr size_t MaxChannels = 6;
+
+    struct Parameters {
+        u32 sample_rate;
+        u32 channel_count;
+        f32 momentary_time_constant;  // Default: 0.4s
+        f32 short_term_time_constant; // Default: 3.0s
+    };
+
+    LoudnessCalculator();
+    ~LoudnessCalculator();
+
+    /**
+     * Initialize the loudness calculator.
+     *
+     * @param params - Configuration parameters.
+     * @return True if initialization succeeded.
+     */
+    bool Initialize(const Parameters& params);
+
+    /**
+     * Finalize and cleanup the calculator.
+     */
+    void Finalize();
+
+    /**
+     * Reset the calculator state.
+     */
+    void Reset();
+
+    /**
+     * Analyze a block of audio samples.
+     *
+     * @param samples - Interleaved audio samples.
+     * @param sample_count - Number of samples per channel.
+     */
+    void Analyze(std::span<const f32> samples, u32 sample_count);
+
+    /**
+     * Get momentary loudness (400ms window).
+     *
+     * @return Loudness in LUFS.
+     */
+    f32 GetMomentaryLoudness() const;
+
+    /**
+     * Get short-term loudness (3s window).
+     *
+     * @return Loudness in LUFS.
+     */
+    f32 GetShortTermLoudness() const;
+
+    /**
+     * Get integrated loudness (entire duration).
+     *
+     * @return Loudness in LUFS.
+     */
+    f32 GetIntegratedLoudness() const;
+
+    /**
+     * Get loudness range.
+     *
+     * @return Loudness range in LU.
+     */
+    f32 GetLoudnessRange() const;
+
+    /**
+     * Set momentary loudness low-pass filter time constant.
+     *
+     * @param time_constant - Time constant in seconds.
+     */
+    void SetMomentaryLoudnessLpfTc(f32 time_constant);
+
+private:
+    struct KWeightingFilter {
+        // Shelf filter coefficients
+        f32 b0_shelf, b1_shelf, b2_shelf;
+        f32 a1_shelf, a2_shelf;
+        f32 z1_shelf[MaxChannels], z2_shelf[MaxChannels];
+
+        // High-pass filter coefficients
+        f32 b0_hp, b1_hp, b2_hp;
+        f32 a1_hp, a2_hp;
+        f32 z1_hp[MaxChannels], z2_hp[MaxChannels];
+    };
+
+    void InitializeKWeightingFilter();
+    f32 ApplyKWeighting(f32 sample, u32 channel);
+    f32 CalculateLoudness(f32 mean_square);
+
+    Parameters params_{};
+    bool initialized_{false};
+
+    KWeightingFilter k_filter_{};
+
+    // Loudness measurements
+    f32 momentary_loudness_{-70.0f};
+    f32 short_term_loudness_{-70.0f};
+    f32 integrated_loudness_{-70.0f};
+    f32 loudness_range_{0.0f};
+
+    // Integrated measurement state
+    f32 integrated_sum_{0.0f};
+    u64 integrated_count_{0};
+
+    // Ring buffers for time-windowed measurements
+    std::array<f32, 48000> momentary_buffer_{};  // 1 second at 48kHz
+    std::array<f32, 144000> short_term_buffer_{}; // 3 seconds at 48kHz
+    size_t buffer_index_{0};
+};
+
+} // namespace AudioCore

src/audio_core/device/audio_buffer_list.h

@@ -0,0 +1,93 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <span>
+
+#include "common/common_types.h"
+
+namespace AudioCore {
+
+template <typename BufferType>
+class AudioBufferList {
+public:
+    static constexpr size_t BufferCount = 32;
+
+    AudioBufferList() = default;
+    ~AudioBufferList() = default;
+
+    void clear() {
+        count = 0;
+        head_index = 0;
+        tail_index = 0;
+    }
+
+    void push_back(const BufferType& buffer) {
+        if (count >= BufferCount) {
+            return;
+        }
+
+        buffers[tail_index] = buffer;
+        tail_index = (tail_index + 1) % BufferCount;
+        count++;
+    }
+
+    void pop_front() {
+        if (count == 0) {
+            return;
+        }
+
+        head_index = (head_index + 1) % BufferCount;
+        count--;
+    }
+
+    BufferType& front() {
+        return buffers[head_index];
+    }
+
+    const BufferType& front() const {
+        return buffers[head_index];
+    }
+
+    BufferType& back() {
+        size_t index = (tail_index + BufferCount - 1) % BufferCount;
+        return buffers[index];
+    }
+
+    const BufferType& back() const {
+        size_t index = (tail_index + BufferCount - 1) % BufferCount;
+        return buffers[index];
+    }
+
+    size_t size() const {
+        return count;
+    }
+
+    bool empty() const {
+        return count == 0;
+    }
+
+    bool full() const {
+        return count >= BufferCount;
+    }
+
+    bool contains(const BufferType* buffer_ptr) const {
+        for (size_t i = 0; i < count; i++) {
+            size_t index = (head_index + i) % BufferCount;
+            if (&buffers[index] == buffer_ptr) {
+                return true;
+            }
+        }
+        return false;
+    }
+
+private:
+    std::array<BufferType, BufferCount> buffers{};
+    size_t count{0};
+    size_t head_index{0};
+    size_t tail_index{0};
+};
+
+} // namespace AudioCore

src/audio_core/device/shared_ring_buffer.cpp

@@ -0,0 +1,169 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/device/shared_ring_buffer.h"
+#include "audio_core/renderer/final_output_recorder/final_output_recorder_buffer.h"
+#include "common/logging/log.h"
+#include "core/memory.h"
+
+namespace AudioCore {
+
+SharedRingBuffer::SharedRingBuffer() = default;
+SharedRingBuffer::~SharedRingBuffer() {
+    Finalize();
+}
+
+bool SharedRingBuffer::Initialize(Core::Memory::Memory& memory_, VAddr buffer_address_,
+                                   u64 buffer_size_, VAddr data_address_, u64 data_size_,
+                                   u32 buffer_count) {
+    if (initialized) {
+        return false;
+    }
+
+    memory = std::addressof(memory_);
+    buffer_address = buffer_address_;
+    buffer_size = buffer_size_;
+    data_address = data_address_;
+    data_size = data_size_;
+    max_buffer_count = buffer_count;
+    current_buffer_count = 0;
+    read_offset = 0;
+    write_offset = 0;
+    initialized = true;
+
+    ClearSharedState();
+    return true;
+}
+
+void SharedRingBuffer::Finalize() {
+    if (!initialized) {
+        return;
+    }
+
+    initialized = false;
+    memory = static_cast<Core::Memory::Memory*>(nullptr);
+    buffer_address = 0;
+    buffer_size = 0;
+    data_address = 0;
+    data_size = 0;
+    max_buffer_count = 0;
+    current_buffer_count = 0;
+    read_offset = 0;
+    write_offset = 0;
+}
+
+void SharedRingBuffer::ClearSharedState() {
+    if (!initialized || !memory) {
+        return;
+    }
+
+    SharedState state{};
+    state.read_offset = 0;
+    state.write_offset = 0;
+    state.buffer_count = 0;
+    state.sample_rate = 48000;
+    state.channel_count = 2;
+    state.sample_format = 2; // PCM16
+
+    memory->WriteBlock(data_address, &state, sizeof(SharedState));
+}
+
+bool SharedRingBuffer::GetReleasedBufferForPlayback(FinalOutputRecorderBuffer& out_buffer) {
+    // Not implemented for playback
+    return false;
+}
+
+bool SharedRingBuffer::GetReleasedBufferForRecord(FinalOutputRecorderBuffer& out_buffer) {
+    if (!initialized || !memory || current_buffer_count == 0) {
+        return false;
+    }
+
+    // Read the buffer info from the ring
+    memory->ReadBlock(buffer_address + read_offset, &out_buffer, sizeof(FinalOutputRecorderBuffer));
+
+    read_offset += sizeof(FinalOutputRecorderBuffer);
+    if (read_offset >= buffer_size) {
+        read_offset = 0;
+    }
+
+    current_buffer_count--;
+    return true;
+}
+
+bool SharedRingBuffer::AppendBufferForPlayback(const FinalOutputRecorderBuffer& buffer) {
+    // Not implemented for playback
+    return false;
+}
+
+bool SharedRingBuffer::HasCapacityForAppend() const {
+    return initialized && current_buffer_count < max_buffer_count;
+}
+
+bool SharedRingBuffer::HasAvailableBuffer() const {
+    return initialized && current_buffer_count > 0;
+}
+
+bool SharedRingBuffer::AppendBufferForRecord(const FinalOutputRecorderBuffer& buffer) {
+    if (!initialized || !memory || !HasCapacityForRecord()) {
+        return false;
+    }
+
+    // Write the buffer info to the ring
+    memory->WriteBlock(buffer_address + write_offset, &buffer,
+                       sizeof(FinalOutputRecorderBuffer));
+
+    write_offset += sizeof(FinalOutputRecorderBuffer);
+    if (write_offset >= buffer_size) {
+        write_offset = 0;
+    }
+
+    current_buffer_count++;
+    return true;
+}
+
+bool SharedRingBuffer::HasCapacityForRecord() const {
+    return initialized && current_buffer_count < max_buffer_count;
+}
+
+bool SharedRingBuffer::ContainsBuffer(VAddr buffer_address_) const {
+    if (!initialized || !memory) {
+        return false;
+    }
+
+    u64 temp_offset = read_offset;
+    for (u32 i = 0; i < current_buffer_count; i++) {
+        FinalOutputRecorderBuffer buffer;
+        memory->ReadBlock(buffer_address + temp_offset, &buffer,
+                          sizeof(FinalOutputRecorderBuffer));
+
+        if (buffer.samples == buffer_address_) {
+            return true;
+        }
+
+        temp_offset += sizeof(FinalOutputRecorderBuffer);
+        if (temp_offset >= buffer_size) {
+            temp_offset = 0;
+        }
+    }
+
+    return false;
+}
+
+u32 SharedRingBuffer::GetBufferCount() const {
+    return current_buffer_count;
+}
+
+u64 SharedRingBuffer::GetSampleProcessedSampleCount() const {
+    // Return the total number of samples processed
+    return 0; // TODO: Track this
+}
+
+u64 SharedRingBuffer::GetWorkBufferDataSizeBytes() const {
+    return buffer_size;
+}
+
+VAddr SharedRingBuffer::GetWorkBufferDataAddress() const {
+    return buffer_address;
+}
+
+} // namespace AudioCore

src/audio_core/device/shared_ring_buffer.h

@@ -0,0 +1,167 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <memory>
+#include <span>
+
+#include "common/common_types.h"
+
+namespace Core::Memory {
+class Memory;
+}
+
+namespace AudioCore {
+
+struct FinalOutputRecorderBuffer;
+
+/**
+ * Shared ring buffer for final output recording.
+ * Manages a circular buffer shared between the emulator and the game.
+ */
+class SharedRingBuffer {
+public:
+    struct SharedState {
+        u64 read_offset;
+        u64 write_offset;
+        u32 buffer_count;
+        u32 sample_rate;
+        u32 channel_count;
+        u32 sample_format;
+    };
+
+    SharedRingBuffer();
+    ~SharedRingBuffer();
+
+    // Prevent copying
+    SharedRingBuffer(const SharedRingBuffer&) = delete;
+    SharedRingBuffer& operator=(const SharedRingBuffer&) = delete;
+
+    /**
+     * Initialize the shared ring buffer.
+     *
+     * @param memory - Core memory for accessing shared memory.
+     * @param buffer_address - Address of the ring buffer data.
+     * @param buffer_size - Size of the ring buffer in bytes.
+     * @param data_address - Address of the shared state.
+     * @param data_size - Size of the shared state.
+     * @param buffer_count - Maximum number of buffers that can be queued.
+     * @return True if initialization succeeded.
+     */
+    bool Initialize(Core::Memory::Memory& memory, VAddr buffer_address, u64 buffer_size,
+                    VAddr data_address, u64 data_size, u32 buffer_count);
+
+    /**
+     * Finalize and cleanup the ring buffer.
+     */
+    void Finalize();
+
+    /**
+     * Clear the shared state.
+     */
+    void ClearSharedState();
+
+    /**
+     * Get a released buffer for playback.
+     *
+     * @param out_buffer - Output buffer information.
+     * @return True if a buffer was available.
+     */
+    bool GetReleasedBufferForPlayback(FinalOutputRecorderBuffer& out_buffer);
+
+    /**
+     * Get a released buffer for recording.
+     *
+     * @param out_buffer - Output buffer information.
+     * @return True if a buffer was available.
+     */
+    bool GetReleasedBufferForRecord(FinalOutputRecorderBuffer& out_buffer);
+
+    /**
+     * Append a buffer for playback.
+     *
+     * @param buffer - Buffer to append.
+     * @return True if the buffer was successfully appended.
+     */
+    bool AppendBufferForPlayback(const FinalOutputRecorderBuffer& buffer);
+
+    /**
+     * Check if there's capacity to append a new buffer.
+     *
+     * @return True if there's capacity.
+     */
+    bool HasCapacityForAppend() const;
+
+    /**
+     * Check if there's an available buffer to retrieve.
+     *
+     * @return True if a buffer is available.
+     */
+    bool HasAvailableBuffer() const;
+
+    /**
+     * Append a buffer for recording.
+     *
+     * @param buffer - Buffer to append.
+     * @return True if the buffer was successfully appended.
+     */
+    bool AppendBufferForRecord(const FinalOutputRecorderBuffer& buffer);
+
+    /**
+     * Check if there's capacity to record a new buffer.
+     *
+     * @return True if there's capacity.
+     */
+    bool HasCapacityForRecord() const;
+
+    /**
+     * Check if the ring buffer contains a specific buffer.
+     *
+     * @param buffer_address - Address of the buffer to check.
+     * @return True if the buffer is in the ring.
+     */
+    bool ContainsBuffer(VAddr buffer_address) const;
+
+    /**
+     * Get the current buffer count.
+     *
+     * @return Number of buffers currently in the ring.
+     */
+    u32 GetBufferCount() const;
+
+    /**
+     * Get the sample processed sample count.
+     *
+     * @return Number of samples processed.
+     */
+    u64 GetSampleProcessedSampleCount() const;
+
+    /**
+     * Get the work buffer data size in bytes.
+     *
+     * @return Size of the work buffer data.
+     */
+    u64 GetWorkBufferDataSizeBytes() const;
+
+    /**
+     * Get the work buffer data address.
+     *
+     * @return Address of the work buffer data.
+     */
+    VAddr GetWorkBufferDataAddress() const;
+
+private:
+    Core::Memory::Memory* memory{nullptr};
+    VAddr buffer_address{0};
+    u64 buffer_size{0};
+    VAddr data_address{0};
+    u64 data_size{0};
+    u32 max_buffer_count{0};
+    u32 current_buffer_count{0};
+    u64 read_offset{0};
+    u64 write_offset{0};
+    bool initialized{false};
+};
+
+} // namespace AudioCore

src/audio_core/in/audio_in_system.cpp

@@ -206,6 +206,16 @@ void System::SetVolume(const f32 volume_) {
     session->SetVolume(volume_);
 }
 
+f32 System::GetDeviceGain() const {
+    return device_gain;
+}
+
+void System::SetDeviceGain(const f32 gain) {
+    device_gain = gain;
+    // Apply the device gain to the session
+    session->SetVolume(volume * device_gain);
+}
+
 bool System::ContainsAudioBuffer(const u64 tag) const {
     return buffers.ContainsBuffer(tag);
 }

src/audio_core/in/audio_in_system.h

@@ -213,6 +213,20 @@ public:
      */
     void SetVolume(f32 volume);
 
+    /**
+     * Get this system's current device gain.
+     *
+     * @return The system's current device gain.
+     */
+    f32 GetDeviceGain() const;
+
+    /**
+     * Set this system's current device gain.
+     *
+     * @param gain The new device gain.
+     */
+    void SetDeviceGain(f32 gain);
+
     /**
      * Does the system contain this buffer?
      *
@@ -269,6 +283,8 @@ private:
     std::string name{};
     /// Volume of this system
     f32 volume{1.0f};
+    /// Device gain of this system
+    f32 device_gain{1.0f};
     /// Is this system's device USB?
     bool is_uac{false};
 };

src/audio_core/renderer/behavior/behavior_info.cpp

@@ -198,4 +198,12 @@ bool BehaviorInfo::IsSplitterPrevVolumeResetSupported() const {
     return CheckFeatureSupported(SupportTags::SplitterPrevVolumeReset, user_revision);
 }
 
+bool BehaviorInfo::IsSplitterDestinationV2bSupported() const {
+    return CheckFeatureSupported(SupportTags::SplitterDestinationV2b, user_revision);
+}
+
+bool BehaviorInfo::IsVoiceInParameterV2Supported() const {
+    return CheckFeatureSupported(SupportTags::VoiceInParameterV2, user_revision);
+}
+
 } // namespace AudioCore::Renderer

src/audio_core/renderer/behavior/behavior_info.h

@@ -378,6 +378,22 @@ public:
      */
     bool IsSplitterPrevVolumeResetSupported() const;
 
+    /**
+     * Check if splitter destination v2b parameter format is supported (revision 15+).
+     * This uses the extended parameter format with biquad filter fields.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsSplitterDestinationV2bSupported() const;
+
+    /**
+     * Check if voice input parameter v2 format is supported (revision 15+).
+     * This uses the extended parameter format with float biquad filters.
+     *
+     * @return True if supported, otherwise false.
+     */
+    bool IsVoiceInParameterV2Supported() const;
+
     /// Host version
     u32 process_revision;
     /// User version

src/audio_core/renderer/behavior/info_updater.cpp

@@ -61,8 +61,6 @@ Result InfoUpdater::UpdateVoices(VoiceContext& voice_context,
     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};
 
@@ -73,8 +71,97 @@ Result InfoUpdater::UpdateVoices(VoiceContext& voice_context,
 
     u32 new_voice_count{0};
 
+    // Two input formats exist: legacy (0x170) and v2 with float biquad (0x188).
+    const bool use_v2 = behaviour.IsVoiceInParameterV2Supported();
+    const u32 in_stride = use_v2 ? 0x188u : static_cast<u32>(sizeof(VoiceInfo::InParameter));
+
     for (u32 i = 0; i < voice_count; i++) {
-        const auto& in_param{in_params[i]};
+        VoiceInfo::InParameter local_in{};
+        // Store original float biquad coefficients for REV15+
+        std::array<VoiceInfo::BiquadFilterParameter2, MaxBiquadFilters> float_biquads{};
+
+        if (!use_v2) {
+            const auto* in_param_ptr = reinterpret_cast<const VoiceInfo::InParameter*>(input + i * sizeof(VoiceInfo::InParameter));
+            local_in = *in_param_ptr;
+        } else {
+            struct VoiceInParameterV2 {
+                u32 id;
+                u32 node_id;
+                bool is_new;
+                bool in_use;
+                PlayState play_state;
+                SampleFormat sample_format;
+                u32 sample_rate;
+                u32 priority;
+                u32 sort_order;
+                u32 channel_count;
+                f32 pitch;
+                f32 volume;
+                // Two BiquadFilterParameter2 (0x18 each) -> ignored/converted
+                struct BiquadV2 { bool enable; u8 r1; u8 r2; u8 r3; std::array<f32,3> b; std::array<f32,2> a; } biquads[2];
+                u32 wave_buffer_count;
+                u32 wave_buffer_index;
+                u32 reserved1;
+                u64 src_data_address;
+                u64 src_data_size;
+                s32 mix_id;
+                u32 splitter_id;
+                std::array<VoiceInfo::WaveBufferInternal, MaxWaveBuffers> wavebuffers;
+                std::array<u32, MaxChannels> channel_resource_ids;
+                bool clear_voice_drop;
+                u8 flush_wave_buffer_count;
+                u16 reserved2;
+                VoiceInfo::Flags flags;
+                SrcQuality src_quality;
+                u32 external_ctx;
+                u32 external_ctx_size;
+                u32 reserved3[2];
+            };
+            const auto* vin = reinterpret_cast<const VoiceInParameterV2*>(input + i * in_stride);
+            local_in.id = vin->id;
+            local_in.node_id = vin->node_id;
+            local_in.is_new = vin->is_new;
+            local_in.in_use = vin->in_use;
+            local_in.play_state = vin->play_state;
+            local_in.sample_format = vin->sample_format;
+            local_in.sample_rate = vin->sample_rate;
+            local_in.priority = static_cast<s32>(vin->priority);
+            local_in.sort_order = static_cast<s32>(vin->sort_order);
+            local_in.channel_count = vin->channel_count;
+            local_in.pitch = vin->pitch;
+            local_in.volume = vin->volume;
+
+            // For REV15+, we keep float coefficients separate and only convert for compatibility
+            for (size_t filter_idx = 0; filter_idx < MaxBiquadFilters; filter_idx++) {
+                const auto& src = vin->biquads[filter_idx];
+                auto& dst = local_in.biquads[filter_idx];
+                dst.enabled = src.enable;
+                // Convert float coefficients to fixed-point Q2.14 for legacy path
+                dst.b[0] = static_cast<s16>(std::clamp(src.b[0] * 16384.0f, -32768.0f, 32767.0f));
+                dst.b[1] = static_cast<s16>(std::clamp(src.b[1] * 16384.0f, -32768.0f, 32767.0f));
+                dst.b[2] = static_cast<s16>(std::clamp(src.b[2] * 16384.0f, -32768.0f, 32767.0f));
+                dst.a[0] = static_cast<s16>(std::clamp(src.a[0] * 16384.0f, -32768.0f, 32767.0f));
+                dst.a[1] = static_cast<s16>(std::clamp(src.a[1] * 16384.0f, -32768.0f, 32767.0f));
+
+                // Also store the native float version
+                float_biquads[filter_idx].enabled = src.enable;
+                float_biquads[filter_idx].numerator = src.b;
+                float_biquads[filter_idx].denominator = src.a;
+            }
+            local_in.wave_buffer_count = vin->wave_buffer_count;
+            local_in.wave_buffer_index = static_cast<u16>(vin->wave_buffer_index);
+            local_in.src_data_address = static_cast<CpuAddr>(vin->src_data_address);
+            local_in.src_data_size = vin->src_data_size;
+            local_in.mix_id = static_cast<u32>(vin->mix_id);
+            local_in.splitter_id = vin->splitter_id;
+            local_in.wave_buffer_internal = vin->wavebuffers;
+            local_in.channel_resource_ids = vin->channel_resource_ids;
+            local_in.clear_voice_drop = vin->clear_voice_drop;
+            local_in.flush_buffer_count = vin->flush_wave_buffer_count;
+            local_in.flags = vin->flags;
+            local_in.src_quality = vin->src_quality;
+        }
+        const auto& in_param = local_in;
         std::array<VoiceState*, MaxChannels> voice_states{};
 
         if (!in_param.in_use) {
@@ -98,6 +185,14 @@ Result InfoUpdater::UpdateVoices(VoiceContext& voice_context,
         BehaviorInfo::ErrorInfo update_error{};
         voice_info.UpdateParameters(update_error, in_param, pool_mapper, behaviour);
 
+        // For REV15+, store the native float biquad coefficients
+        if (use_v2) {
+            voice_info.use_float_biquads = true;
+            voice_info.biquads_float = float_biquads;
+        } else {
+            voice_info.use_float_biquads = false;
+        }
+
         if (!update_error.error_code.IsSuccess()) {
             behaviour.AppendError(update_error);
         }
@@ -118,7 +213,7 @@ Result InfoUpdater::UpdateVoices(VoiceContext& voice_context,
         new_voice_count += in_param.channel_count;
     }
 
-    auto consumed_input_size{voice_count * static_cast<u32>(sizeof(VoiceInfo::InParameter))};
+    auto consumed_input_size{voice_count * in_stride};
     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={}",
@@ -254,18 +349,29 @@ Result InfoUpdater::UpdateMixes(MixContext& mix_context, const u32 mix_buffer_co
                                 EffectContext& effect_context, SplitterContext& splitter_context) {
     s32 mix_count{0};
     u32 consumed_input_size{0};
+    u32 input_mix_size{0};
 
     if (behaviour.IsMixInParameterDirtyOnlyUpdateSupported()) {
         auto in_dirty_params{reinterpret_cast<const MixInfo::InDirtyParameter*>(input)};
         mix_count = in_dirty_params->count;
+
+        // Validate against expected header size to ensure structure is correct
+        if (mix_count < 0 || mix_count > 0x100) {
+            LOG_ERROR(Service_Audio,
+                "Invalid mix count from dirty parameter: count={}, magic=0x{:X}, expected_size={}",
+                mix_count, in_dirty_params->magic, in_header->mix_size);
+            return Service::Audio::ResultInvalidUpdateInfo;
+        }
+
+        consumed_input_size += static_cast<u32>(sizeof(MixInfo::InDirtyParameter));
         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));
     }
 
+    input_mix_size = static_cast<u32>(mix_count * sizeof(MixInfo::InParameter));
+    consumed_input_size += input_mix_size;
+
     if (mix_buffer_count == 0) {
         return Service::Audio::ResultInvalidUpdateInfo;
     }
@@ -330,7 +436,7 @@ Result InfoUpdater::UpdateMixes(MixContext& mix_context, const u32 mix_buffer_co
         return Service::Audio::ResultInvalidUpdateInfo;
     }
 
-    input += mix_count * sizeof(MixInfo::InParameter);
+    input += input_mix_size;
 
     return ResultSuccess;
 }

src/audio_core/renderer/command/command_buffer.cpp

@@ -234,6 +234,14 @@ void CommandBuffer::GenerateBiquadFilterCommand(const s32 node_id, VoiceInfo& vo
 
     cmd.biquad = voice_info.biquads[biquad_index];
 
+    // REV15+: Use native float coefficients if available
+    if (voice_info.use_float_biquads) {
+        cmd.biquad_float = voice_info.biquads_float[biquad_index];
+        cmd.use_float_coefficients = true;
+    } else {
+        cmd.use_float_coefficients = false;
+    }
+
     cmd.state = memory_pool->Translate(CpuAddr(voice_state.biquad_states[biquad_index].data()),
                                        MaxBiquadFilters * sizeof(VoiceState::BiquadFilterState));
 
@@ -260,6 +268,9 @@ void CommandBuffer::GenerateBiquadFilterCommand(const s32 node_id, EffectInfoBas
     cmd.biquad.b = parameter.b;
     cmd.biquad.a = parameter.a;
 
+    // Effects use legacy fixed-point format
+    cmd.use_float_coefficients = false;
+
     cmd.state = memory_pool->Translate(CpuAddr(state),
                                        MaxBiquadFilters * sizeof(VoiceState::BiquadFilterState));
 
@@ -655,6 +666,14 @@ void CommandBuffer::GenerateMultitapBiquadFilterCommand(const s32 node_id, Voice
     cmd.output = buffer_count + channel;
     cmd.biquads = voice_info.biquads;
 
+    // REV15+: Use native float coefficients if available
+    if (voice_info.use_float_biquads) {
+        cmd.biquads_float = voice_info.biquads_float;
+        cmd.use_float_coefficients = true;
+    } else {
+        cmd.use_float_coefficients = false;
+    }
+
     cmd.states[0] =
         memory_pool->Translate(CpuAddr(voice_state.biquad_states[0].data()),
                                MaxBiquadFilters * sizeof(VoiceState::BiquadFilterState));

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

@@ -48,6 +48,39 @@ void ApplyBiquadFilterFloat(std::span<s32> output, std::span<const s32> input,
     state.s3 = Common::BitCast<s64>(s[3]);
 }
 
+/**
+ * Biquad filter float implementation with native float coefficients (SDK REV15+).
+ */
+void ApplyBiquadFilterFloat2(std::span<s32> output, std::span<const s32> input,
+                             std::array<f32, 3>& b, std::array<f32, 2>& a,
+                             VoiceState::BiquadFilterState& state, const u32 sample_count) {
+    constexpr f64 min{std::numeric_limits<s32>::min()};
+    constexpr f64 max{std::numeric_limits<s32>::max()};
+
+    std::array<f64, 3> b_double{static_cast<f64>(b[0]), static_cast<f64>(b[1]), static_cast<f64>(b[2])};
+    std::array<f64, 2> a_double{static_cast<f64>(a[0]), static_cast<f64>(a[1])};
+    std::array<f64, 4> s{Common::BitCast<f64>(state.s0), Common::BitCast<f64>(state.s1),
+                         Common::BitCast<f64>(state.s2), Common::BitCast<f64>(state.s3)};
+
+    for (u32 i = 0; i < sample_count; i++) {
+        f64 in_sample{static_cast<f64>(input[i])};
+        auto sample{in_sample * b_double[0] + s[0] * b_double[1] + s[1] * b_double[2] +
+                    s[2] * a_double[0] + s[3] * a_double[1]};
+
+        output[i] = static_cast<s32>(std::clamp(sample, min, max));
+
+        s[1] = s[0];
+        s[0] = in_sample;
+        s[3] = s[2];
+        s[2] = sample;
+    }
+
+    state.s0 = Common::BitCast<s64>(s[0]);
+    state.s1 = Common::BitCast<s64>(s[1]);
+    state.s2 = Common::BitCast<s64>(s[2]);
+    state.s3 = Common::BitCast<s64>(s[3]);
+}
+
 /**
  * Biquad filter s32 implementation.
  *
@@ -95,8 +128,14 @@ void BiquadFilterCommand::Process(const AudioRenderer::CommandListProcessor& pro
         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_,
+        // REV15+: Use native float coefficients if available
-                               processor.sample_count);
+        if (use_float_coefficients) {
+            ApplyBiquadFilterFloat2(output_buffer, input_buffer, biquad_float.numerator,
+                                   biquad_float.denominator, *state_, processor.sample_count);
+        } else {
+            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);

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

@@ -48,14 +48,18 @@ struct BiquadFilterCommand : ICommand {
     s16 input;
     /// Output mix buffer index
     s16 output;
-    /// Input parameters for biquad
+    /// Input parameters for biquad (legacy fixed-point)
     VoiceInfo::BiquadFilterParameter biquad;
+    /// Input parameters for biquad (REV15+ native float)
+    VoiceInfo::BiquadFilterParameter2 biquad_float;
     /// 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;
+    /// If true, use native float coefficients (REV15+)
+    bool use_float_coefficients;
 };
 
 /**
@@ -72,4 +76,18 @@ 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);
 
+/**
+ * Biquad filter float implementation with native float coefficients (SDK REV15+).
+ *
+ * @param output       - Output container for filtered samples.
+ * @param input        - Input container for samples to be filtered.
+ * @param b            - Feedforward coefficients (float).
+ * @param a            - Feedback coefficients (float).
+ * @param state        - State to track previous samples.
+ * @param sample_count - Number of samples to process.
+ */
+void ApplyBiquadFilterFloat2(std::span<s32> output, std::span<const s32> input,
+                             std::array<f32, 3>& b, std::array<f32, 2>& a,
+                             VoiceState::BiquadFilterState& state, const u32 sample_count);
+
 } // namespace AudioCore::Renderer

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

@@ -0,0 +1,90 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <algorithm>
+#include <cmath>
+#include <span>
+
+#include "audio_core/adsp/apps/audio_renderer/command_list_processor.h"
+#include "audio_core/renderer/command/effect/limiter.h"
+#include "core/memory.h"
+
+namespace AudioCore::Renderer {
+
+void LimiterCommand::Dump([[maybe_unused]] const AudioRenderer::CommandListProcessor& processor,
+                          std::string& string) {
+    string += fmt::format("LimiterCommand\n\tenabled {} channels {}\n", effect_enabled,
+                          parameter.channel_count);
+}
+
+void LimiterCommand::Process(const AudioRenderer::CommandListProcessor& processor) {
+    std::array<std::span<const s32>, MaxChannels> input_buffers{};
+    std::array<std::span<s32>, MaxChannels> output_buffers{};
+
+    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_buffer{reinterpret_cast<LimiterInfo::State*>(state)};
+
+    if (effect_enabled) {
+        // Convert parameters
+        const f32 attack_coeff =
+            std::exp(-1.0f / (parameter.attack_time * processor.target_sample_rate / 1000.0f));
+        const f32 release_coeff =
+            std::exp(-1.0f / (parameter.release_time * processor.target_sample_rate / 1000.0f));
+        const f32 threshold_linear = std::pow(10.0f, parameter.threshold / 20.0f);
+        const f32 makeup_gain_linear = std::pow(10.0f, parameter.makeup_gain / 20.0f);
+
+        for (u32 sample = 0; sample < processor.sample_count; sample++) {
+            // Find peak across all channels
+            f32 peak = 0.0f;
+            for (u32 ch = 0; ch < parameter.channel_count; ch++) {
+                const f32 abs_sample = std::abs(static_cast<f32>(input_buffers[ch][sample]));
+                peak = std::max(peak, abs_sample);
+            }
+
+            // Update envelope
+            if (peak > state_buffer->envelope) {
+                state_buffer->envelope =
+                    attack_coeff * state_buffer->envelope + (1.0f - attack_coeff) * peak;
+            } else {
+                state_buffer->envelope =
+                    release_coeff * state_buffer->envelope + (1.0f - release_coeff) * peak;
+            }
+
+            // Calculate gain reduction
+            f32 gain = 1.0f;
+            if (state_buffer->envelope > threshold_linear) {
+                const f32 over = state_buffer->envelope / threshold_linear;
+                gain = 1.0f / std::pow(over, (parameter.ratio - 1.0f) / parameter.ratio);
+            }
+
+            state_buffer->gain_reduction = gain;
+
+            // Apply limiting with makeup gain
+            const f32 total_gain = gain * makeup_gain_linear;
+            for (u32 ch = 0; ch < parameter.channel_count; ch++) {
+                output_buffers[ch][sample] =
+                    static_cast<s32>(input_buffers[ch][sample] * total_gain);
+            }
+        }
+    } else {
+        // Bypass: just copy input to output
+        for (u32 ch = 0; ch < parameter.channel_count; ch++) {
+            if (inputs[ch] != outputs[ch]) {
+                std::memcpy(output_buffers[ch].data(), input_buffers[ch].data(),
+                            output_buffers[ch].size_bytes());
+            }
+        }
+    }
+}
+
+bool LimiterCommand::Verify(const AudioRenderer::CommandListProcessor& processor) {
+    return true;
+}
+
+} // namespace AudioCore::Renderer

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

@@ -0,0 +1,60 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron 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/limiter.h"
+#include "common/common_types.h"
+
+namespace AudioCore::ADSP::AudioRenderer {
+class CommandListProcessor;
+}
+
+namespace AudioCore::Renderer {
+
+/**
+ * AudioRenderer command for limiting volume with attack/release controls.
+ */
+struct LimiterCommand : 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 AudioRenderer::CommandListProcessor& processor, std::string& string) override;
+
+    /**
+     * Process this command.
+     *
+     * @param processor - The CommandListProcessor processing this command.
+     */
+    void Process(const AudioRenderer::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 AudioRenderer::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
+    LimiterInfo::ParameterVersion2 parameter;
+    /// State, updated each call
+    CpuAddr state;
+    /// Game-supplied workbuffer (Unused)
+    CpuAddr workbuffer;
+    /// Is this effect enabled?
+    bool effect_enabled;
+};
+
+} // namespace AudioCore::Renderer

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

@@ -33,8 +33,14 @@ void MultiTapBiquadFilterCommand::Process(const AudioRenderer::CommandListProces
             *state = {};
         }
 
-        ApplyBiquadFilterFloat(output_buffer, input_buffer, biquads[i].b, biquads[i].a, *state,
+        // REV15+: Use native float coefficients if available
-                               processor.sample_count);
+        if (use_float_coefficients) {
+            ApplyBiquadFilterFloat2(output_buffer, input_buffer, biquads_float[i].numerator,
+                                   biquads_float[i].denominator, *state, processor.sample_count);
+        } else {
+            ApplyBiquadFilterFloat(output_buffer, input_buffer, biquads[i].b, biquads[i].a, *state,
+                                   processor.sample_count);
+        }
     }
 }
 

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

@@ -47,14 +47,18 @@ struct MultiTapBiquadFilterCommand : ICommand {
     s16 input;
     /// Output mix buffer index
     s16 output;
-    /// Biquad parameters
+    /// Biquad parameters (legacy fixed-point)
     std::array<VoiceInfo::BiquadFilterParameter, MaxBiquadFilters> biquads;
+    /// Biquad parameters (REV15+ native float)
+    std::array<VoiceInfo::BiquadFilterParameter2, MaxBiquadFilters> biquads_float;
     /// 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;
+    /// If true, use native float coefficients (REV15+)
+    bool use_float_coefficients;
 };
 
 } // namespace AudioCore::Renderer

src/audio_core/renderer/effect/limiter.cpp

@@ -0,0 +1,67 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/effect/limiter.h"
+#include "core/hle/result.h"
+
+namespace AudioCore::Renderer {
+
+void LimiterInfo::Update(BehaviorInfo::ErrorInfo& error_info,
+                          const EffectInfoBase::InParameterVersion1& in_params,
+                          const PoolMapper& pool_mapper) {
+    auto in_specific{reinterpret_cast<const ParameterVersion1*>(in_params.specific.data())};
+    auto params{reinterpret_cast<ParameterVersion1*>(parameter.data())};
+
+    std::memcpy(params, in_specific, sizeof(ParameterVersion1));
+    mix_id = in_params.mix_id;
+    process_order = in_params.process_order;
+    enabled = in_params.enabled;
+
+    error_info.error_code = ResultSuccess;
+    error_info.address = CpuAddr(0);
+}
+
+void LimiterInfo::Update(BehaviorInfo::ErrorInfo& error_info,
+                          const EffectInfoBase::InParameterVersion2& in_params,
+                          const PoolMapper& pool_mapper) {
+    auto in_specific{reinterpret_cast<const ParameterVersion2*>(in_params.specific.data())};
+    auto params{reinterpret_cast<ParameterVersion2*>(parameter.data())};
+
+    std::memcpy(params, in_specific, sizeof(ParameterVersion2));
+    mix_id = in_params.mix_id;
+    process_order = in_params.process_order;
+    enabled = in_params.enabled;
+
+    error_info.error_code = ResultSuccess;
+    error_info.address = CpuAddr(0);
+}
+
+void LimiterInfo::UpdateForCommandGeneration() {
+    if (enabled) {
+        usage_state = UsageState::Enabled;
+    } else {
+        usage_state = UsageState::Disabled;
+    }
+
+    auto params{reinterpret_cast<ParameterVersion1*>(parameter.data())};
+    params->state = ParameterState::Updated;
+}
+
+void LimiterInfo::InitializeResultState(EffectResultState& result_state) {
+    auto limiter_state{reinterpret_cast<State*>(result_state.state.data())};
+    limiter_state->envelope = 1.0f;
+    limiter_state->gain_reduction = 1.0f;
+    limiter_state->peak_hold = 0.0f;
+    limiter_state->peak_hold_count = 0;
+    limiter_state->channel_peaks.fill(0.0f);
+}
+
+void LimiterInfo::UpdateResultState(EffectResultState& cpu_state, EffectResultState& dsp_state) {
+    cpu_state = dsp_state;
+}
+
+CpuAddr LimiterInfo::GetWorkbuffer(s32 index) {
+    return GetSingleBuffer(index);
+}
+
+} // namespace AudioCore::Renderer

src/audio_core/renderer/effect/limiter.h

@@ -0,0 +1,101 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+
+#include "audio_core/common/common.h"
+#include "audio_core/renderer/effect/effect_info_base.h"
+#include "common/common_types.h"
+
+namespace AudioCore::Renderer {
+
+/**
+ * A full-featured limiter effect with attack, release, and threshold controls.
+ * More sophisticated than LightLimiter.
+ */
+class LimiterInfo : public EffectInfoBase {
+public:
+    struct ParameterVersion1 {
+        /* 0x00 */ std::array<s8, MaxChannels> inputs;
+        /* 0x06 */ std::array<s8, MaxChannels> outputs;
+        /* 0x0C */ u16 channel_count;
+        /* 0x0E */ u16 padding;
+        /* 0x10 */ s32 sample_rate;
+        /* 0x14 */ f32 attack_time;    // Attack time in milliseconds
+        /* 0x18 */ f32 release_time;   // Release time in milliseconds
+        /* 0x1C */ f32 threshold;      // Threshold in dB
+        /* 0x20 */ f32 makeup_gain;    // Makeup gain in dB
+        /* 0x24 */ f32 ratio;          // Compression ratio
+        /* 0x28 */ ParameterState state;
+        /* 0x29 */ bool is_enabled;
+        /* 0x2A */ char unk2A[0x2];
+    };
+    static_assert(sizeof(ParameterVersion1) <= sizeof(EffectInfoBase::InParameterVersion1),
+                  "LimiterInfo::ParameterVersion1 has the wrong size!");
+
+    using ParameterVersion2 = ParameterVersion1;
+
+    struct State {
+        /* 0x00 */ f32 envelope;
+        /* 0x04 */ f32 gain_reduction;
+        /* 0x08 */ f32 peak_hold;
+        /* 0x0C */ u32 peak_hold_count;
+        /* 0x10 */ std::array<f32, MaxChannels> channel_peaks;
+    };
+    static_assert(sizeof(State) <= sizeof(EffectInfoBase::State),
+                  "LimiterInfo::State is too large!");
+
+    /**
+     * Update the info with new parameters.
+     *
+     * @param error_info - Output error information.
+     * @param in_params - Input parameters.
+     * @param pool_mapper - Memory pool mapper for buffers.
+     */
+    void Update(BehaviorInfo::ErrorInfo& error_info,
+                const EffectInfoBase::InParameterVersion1& in_params,
+                const PoolMapper& pool_mapper);
+
+    /**
+     * Update the info with new parameters (version 2).
+     *
+     * @param error_info - Output error information.
+     * @param in_params - Input parameters.
+     * @param pool_mapper - Memory pool mapper for buffers.
+     */
+    void Update(BehaviorInfo::ErrorInfo& error_info,
+                const EffectInfoBase::InParameterVersion2& in_params,
+                const PoolMapper& pool_mapper);
+
+    /**
+     * Update the usage state for command generation.
+     */
+    void UpdateForCommandGeneration();
+
+    /**
+     * Initialize the result state.
+     *
+     * @param result_state - Result state to initialize.
+     */
+    void InitializeResultState(EffectResultState& result_state);
+
+    /**
+     * Update the result state.
+     *
+     * @param cpu_state - CPU-side result state.
+     * @param dsp_state - DSP-side result state.
+     */
+    void UpdateResultState(EffectResultState& cpu_state, EffectResultState& dsp_state);
+
+    /**
+     * Get a workbuffer address.
+     *
+     * @param index - Index of the workbuffer.
+     * @return Address of the workbuffer.
+     */
+    CpuAddr GetWorkbuffer(s32 index);
+};
+
+} // namespace AudioCore::Renderer

src/audio_core/renderer/final_output_recorder/final_output_recorder_buffer.h

@@ -0,0 +1,39 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "common/common_types.h"
+#include "common/swap.h"
+
+namespace AudioCore {
+
+struct FinalOutputRecorderBuffer {
+    /* 0x00 */ FinalOutputRecorderBuffer* next;
+    /* 0x08 */ VAddr samples;
+    /* 0x10 */ u64 capacity;
+    /* 0x18 */ u64 size;
+    /* 0x20 */ u64 offset;
+    /* 0x28 */ u64 end_timestamp;
+};
+static_assert(sizeof(FinalOutputRecorderBuffer) == 0x30,
+              "FinalOutputRecorderBuffer is an invalid size");
+
+struct FinalOutputRecorderParameter {
+    /* 0x0 */ s32_le sample_rate;
+    /* 0x4 */ u16_le channel_count;
+    /* 0x6 */ u16_le reserved;
+};
+static_assert(sizeof(FinalOutputRecorderParameter) == 0x8,
+              "FinalOutputRecorderParameter is an invalid size");
+
+struct FinalOutputRecorderParameterInternal {
+    /* 0x0 */ u32_le sample_rate;
+    /* 0x4 */ u32_le channel_count;
+    /* 0x8 */ u32_le sample_format;
+    /* 0xC */ u32_le state;
+};
+static_assert(sizeof(FinalOutputRecorderParameterInternal) == 0x10,
+              "FinalOutputRecorderParameterInternal is an invalid size");
+
+} // namespace AudioCore

src/audio_core/renderer/final_output_recorder/final_output_recorder_system.cpp

@@ -0,0 +1,128 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/audio_core.h"
+#include "audio_core/renderer/final_output_recorder/final_output_recorder_system.h"
+#include "audio_core/sink/sink.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/hle/kernel/k_event.h"
+#include "core/hle/result.h"
+#include "core/memory.h"
+
+namespace AudioCore::FinalOutputRecorder {
+
+System::System(Core::System& system_, Kernel::KEvent* event_, size_t session_id_)
+    : system{system_}, buffer_event{event_}, session_id{session_id_} {}
+
+System::~System() = default;
+
+Result System::Initialize(const FinalOutputRecorderParameter& params, Kernel::KProcess* handle_,
+                           u64 applet_resource_user_id_) {
+    handle = handle_;
+    applet_resource_user_id = applet_resource_user_id_;
+
+    sample_rate = TargetSampleRate;
+    sample_format = SampleFormat::PcmInt16;
+    channel_count = params.channel_count <= 2 ? 2 : 6;
+
+    buffers.clear();
+    state = State::Stopped;
+
+    return ResultSuccess;
+}
+
+Result System::Start() {
+    if (state != State::Stopped) {
+        return Service::Audio::ResultOperationFailed;
+    }
+
+    state = State::Started;
+    return ResultSuccess;
+}
+
+Result System::Stop() {
+    if (state == State::Started) {
+        state = State::Stopped;
+        buffers.clear();
+        if (buffer_event) {
+            buffer_event->Signal();
+        }
+    }
+
+    return ResultSuccess;
+}
+
+bool System::AppendBuffer(const FinalOutputRecorderBuffer& buffer, u64 tag) {
+    if (buffers.full()) {
+        return false;
+    }
+
+    auto buffer_copy = buffer;
+    buffers.push_back(buffer_copy);
+
+    if (state == State::Started) {
+        ring_buffer.AppendBufferForRecord(buffer_copy);
+    }
+
+    return true;
+}
+
+void System::ReleaseAndRegisterBuffers() {
+    // Release completed buffers
+    while (ring_buffer.HasAvailableBuffer()) {
+        FinalOutputRecorderBuffer buffer;
+        if (ring_buffer.GetReleasedBufferForRecord(buffer)) {
+            if (buffer_event) {
+                buffer_event->Signal();
+            }
+        }
+    }
+}
+
+bool System::FlushAudioBuffers() {
+    buffers.clear();
+    return true;
+}
+
+u32 System::GetReleasedBuffers(std::span<u64> tags) {
+    u32 released = 0;
+
+    while (ring_buffer.HasAvailableBuffer() && released < tags.size()) {
+        FinalOutputRecorderBuffer buffer;
+        if (ring_buffer.GetReleasedBufferForRecord(buffer)) {
+            tags[released] = buffer.offset; // Use offset as tag
+            released++;
+        } else {
+            break;
+        }
+    }
+
+    return released;
+}
+
+bool System::ContainsBuffer(VAddr buffer_address) const {
+    return ring_buffer.ContainsBuffer(buffer_address);
+}
+
+u64 System::GetBufferEndTime() const {
+    // Return the timestamp of the last recorded sample
+    return system.CoreTiming().GetClockTicks();
+}
+
+Result System::AttachWorkBuffer(VAddr work_buffer, u64 work_buffer_size_) {
+    if (work_buffer == 0 || work_buffer_size_ == 0) {
+        return Service::Audio::ResultInvalidHandle;
+    }
+
+    work_buffer_address = work_buffer;
+    work_buffer_size = work_buffer_size_;
+
+    // Initialize the ring buffer with the work buffer
+    auto& memory = system.ApplicationMemory();
+    ring_buffer.Initialize(memory, work_buffer, work_buffer_size, work_buffer, 0x100, 32);
+
+    return ResultSuccess;
+}
+
+} // namespace AudioCore::FinalOutputRecorder

src/audio_core/renderer/final_output_recorder/final_output_recorder_system.h

@@ -0,0 +1,197 @@
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <memory>
+#include <span>
+#include <string>
+
+#include "audio_core/common/common.h"
+#include "audio_core/device/audio_buffer_list.h"
+#include "audio_core/device/device_session.h"
+#include "audio_core/device/shared_ring_buffer.h"
+#include "audio_core/renderer/final_output_recorder/final_output_recorder_buffer.h"
+#include "core/hle/result.h"
+
+namespace Core {
+class System;
+}
+
+namespace Kernel {
+class KEvent;
+class KProcess;
+} // namespace Kernel
+
+namespace AudioCore::FinalOutputRecorder {
+
+constexpr SessionTypes SessionType = SessionTypes::FinalOutputRecorder;
+
+enum class State {
+    Started,
+    Stopped,
+};
+
+/**
+ * Controls and drives final output recording.
+ */
+class System {
+public:
+    explicit System(Core::System& system, Kernel::KEvent* event, size_t session_id);
+    ~System();
+
+    /**
+     * Initialize the final output recorder.
+     *
+     * @param params                  - Input parameters for the recorder.
+     * @param handle_                 - Process handle for memory access.
+     * @param applet_resource_user_id_ - Applet resource user ID.
+     * @return Result code.
+     */
+    Result Initialize(const FinalOutputRecorderParameter& params, Kernel::KProcess* handle_,
+                      u64 applet_resource_user_id_);
+
+    /**
+     * Start the recorder.
+     *
+     * @return Result code.
+     */
+    Result Start();
+
+    /**
+     * Stop the recorder.
+     *
+     * @return Result code.
+     */
+    Result Stop();
+
+    /**
+     * Append a buffer for recording.
+     *
+     * @param buffer - Buffer to append.
+     * @param tag    - User-defined tag for this buffer.
+     * @return True if the buffer was appended successfully.
+     */
+    bool AppendBuffer(const FinalOutputRecorderBuffer& buffer, u64 tag);
+
+    /**
+     * Release and register buffers.
+     */
+    void ReleaseAndRegisterBuffers();
+
+    /**
+     * Flush all audio buffers.
+     *
+     * @return True if buffers were flushed successfully.
+     */
+    bool FlushAudioBuffers();
+
+    /**
+     * Get released buffers.
+     *
+     * @param tags - Output span to receive buffer tags.
+     * @return Number of buffers released.
+     */
+    u32 GetReleasedBuffers(std::span<u64> tags);
+
+    /**
+     * Check if a buffer is contained in the queue.
+     *
+     * @param buffer_address - Address of the buffer to check.
+     * @return True if the buffer is in the queue.
+     */
+    bool ContainsBuffer(VAddr buffer_address) const;
+
+    /**
+     * Get the current state.
+     *
+     * @return Current recorder state.
+     */
+    State GetState() const {
+        return state;
+    }
+
+    /**
+     * Get the sample rate.
+     *
+     * @return Sample rate in Hz.
+     */
+    u32 GetSampleRate() const {
+        return sample_rate;
+    }
+
+    /**
+     * Get the channel count.
+     *
+     * @return Number of channels.
+     */
+    u32 GetChannelCount() const {
+        return channel_count;
+    }
+
+    /**
+     * Get the sample format.
+     *
+     * @return Sample format.
+     */
+    SampleFormat GetSampleFormat() const {
+        return sample_format;
+    }
+
+    /**
+     * Get the session ID.
+     *
+     * @return Session ID.
+     */
+    size_t GetSessionId() const {
+        return session_id;
+    }
+
+    /**
+     * Get the buffer end timestamp.
+     *
+     * @return End timestamp.
+     */
+    u64 GetBufferEndTime() const;
+
+    /**
+     * Attach work buffer.
+     *
+     * @param work_buffer - Work buffer address.
+     * @param work_buffer_size - Work buffer size.
+     * @return Result code.
+     */
+    Result AttachWorkBuffer(VAddr work_buffer, u64 work_buffer_size);
+
+private:
+    /// Core system
+    Core::System& system;
+    /// Buffer event, signalled when a buffer is ready
+    Kernel::KEvent* buffer_event;
+    /// Session ID of this recorder
+    size_t session_id;
+    /// Device session for output
+    std::unique_ptr<DeviceSession> session;
+    /// Audio buffers
+    AudioBufferList<FinalOutputRecorderBuffer> buffers;
+    /// Shared ring buffer for recording
+    SharedRingBuffer ring_buffer;
+    /// Process handle for memory access
+    Kernel::KProcess* handle;
+    /// Applet resource user ID
+    u64 applet_resource_user_id;
+    /// Sample rate
+    u32 sample_rate{TargetSampleRate};
+    /// Channel count
+    u32 channel_count{2};
+    /// Sample format
+    SampleFormat sample_format{SampleFormat::PcmInt16};
+    /// Current state
+    State state{State::Stopped};
+    /// Work buffer address
+    VAddr work_buffer_address{0};
+    /// Work buffer size
+    u64 work_buffer_size{0};
+};
+
+} // namespace AudioCore::FinalOutputRecorder

src/audio_core/renderer/splitter/splitter_context.cpp

@@ -40,6 +40,7 @@ void SplitterContext::Setup(std::span<SplitterInfo> splitter_infos_, const u32 s
     destinations_count = destination_count_;
     splitter_bug_fixed = splitter_bug_fixed_;
     splitter_prev_volume_reset_supported = behavior.IsSplitterPrevVolumeResetSupported();
+    splitter_float_coeff_supported = behavior.IsSplitterDestinationV2bSupported();
 }
 
 bool SplitterContext::UsingSplitter() const {
@@ -136,25 +137,57 @@ u32 SplitterContext::UpdateInfo(const u8* input, u32 offset, const u32 splitter_
 
 u32 SplitterContext::UpdateData(const u8* input, u32 offset, const u32 count) {
     for (u32 i = 0; i < count; i++) {
-        auto data_header{
+        // Version selection based on float coeff/biquad v2b support.
-            reinterpret_cast<const SplitterDestinationData::InParameter*>(input + offset)};
+        if (!splitter_float_coeff_supported) {
+            const auto* data_header =
+                reinterpret_cast<const SplitterDestinationData::InParameter*>(input + offset);
 
-        if (data_header->magic != GetSplitterSendDataMagic()) {
+            if (data_header->magic != GetSplitterSendDataMagic()) {
-            continue;
+                continue;
+            }
+            if (data_header->id < 0 || data_header->id > destinations_count) {
+                continue;
+            }
+
+            auto modified_params = *data_header;
+            if (!splitter_prev_volume_reset_supported) {
+                modified_params.reset_prev_volume = false;
+            }
+            splitter_destinations[data_header->id].Update(modified_params);
+            offset += sizeof(SplitterDestinationData::InParameter);
+        } else {
+            // Version 2b: struct contains extra biquad filter fields
+            const auto* data_header_v2b =
+                reinterpret_cast<const SplitterDestinationData::InParameterVersion2b*>(input + offset);
+
+            if (data_header_v2b->magic != GetSplitterSendDataMagic()) {
+                continue;
+            }
+            if (data_header_v2b->id < 0 || data_header_v2b->id > destinations_count) {
+                continue;
+            }
+
+            // Map common fields to the old format
+            SplitterDestinationData::InParameter mapped{};
+            mapped.magic = data_header_v2b->magic;
+            mapped.id = data_header_v2b->id;
+            mapped.mix_volumes = data_header_v2b->mix_volumes;
+            mapped.mix_id = data_header_v2b->mix_id;
+            mapped.in_use = data_header_v2b->in_use;
+            mapped.reset_prev_volume = splitter_prev_volume_reset_supported ? data_header_v2b->reset_prev_volume : false;
+
+            // Store biquad filters from V2b (REV15+)
+            auto& destination = splitter_destinations[data_header_v2b->id];
+            destination.Update(mapped);
+
+            // Copy biquad filter parameters
+            auto biquad_filters = destination.GetBiquadFilters();
+            for (size_t filter_idx = 0; filter_idx < MaxBiquadFilters; filter_idx++) {
+                biquad_filters[filter_idx] = data_header_v2b->biquad_filters[filter_idx];
+            }
+
+            offset += static_cast<u32>(sizeof(SplitterDestinationData::InParameterVersion2b));
         }
-
-        if (data_header->id < 0 || data_header->id > destinations_count) {
-            continue;
-        }
-
-        // Create a modified parameter that respects the behavior support
-        auto modified_params = *data_header;
-        if (!splitter_prev_volume_reset_supported) {
-            modified_params.reset_prev_volume = false;
-        }
-
-        splitter_destinations[data_header->id].Update(modified_params);
-        offset += sizeof(SplitterDestinationData::InParameter);
     }
 
     return offset;

src/audio_core/renderer/splitter/splitter_context.h

@@ -186,6 +186,8 @@ private:
     bool splitter_bug_fixed{};
     /// Is explicit previous mix volume reset supported?
     bool splitter_prev_volume_reset_supported{};
+    /// Is float coefficient/biquad filter v2b parameter supported?
+    bool splitter_float_coeff_supported{};
 };
 
 } // namespace Renderer

src/audio_core/renderer/splitter/splitter_destinations_data.cpp

@@ -87,4 +87,12 @@ void SplitterDestinationData::SetNext(SplitterDestinationData* next_) {
     next = next_;
 }
 
+std::span<SplitterDestinationData::BiquadFilterParameter2> SplitterDestinationData::GetBiquadFilters() {
+    return biquad_filters;
+}
+
+std::span<const SplitterDestinationData::BiquadFilterParameter2> SplitterDestinationData::GetBiquadFilters() const {
+    return biquad_filters;
+}
+
 } // namespace AudioCore::Renderer

src/audio_core/renderer/splitter/splitter_destinations_data.h

@@ -10,12 +10,31 @@
 #include "common/common_types.h"
 
 namespace AudioCore::Renderer {
+
+// Forward declaration
+class VoiceInfo;
+
 /**
  * Represents a mixing node, can be connected to a previous and next destination forming a chain
  * that a certain mix buffer will pass through to output.
  */
 class SplitterDestinationData {
 public:
+    /**
+     * Biquad filter parameter with float coefficients (SDK REV15+).
+     * Defined here to avoid circular dependency with VoiceInfo.
+     */
+    struct BiquadFilterParameter2 {
+        /* 0x00 */ bool enabled;
+        /* 0x01 */ u8 reserved1;
+        /* 0x02 */ u8 reserved2;
+        /* 0x03 */ u8 reserved3;
+        /* 0x04 */ std::array<f32, 3> numerator;   // b0, b1, b2
+        /* 0x10 */ std::array<f32, 2> denominator; // a1, a2 (a0 = 1)
+    };
+    static_assert(sizeof(BiquadFilterParameter2) == 0x18,
+                  "BiquadFilterParameter2 has the wrong size!");
+
     struct InParameter {
         /* 0x00 */ u32 magic; // 'SNDD'
         /* 0x04 */ s32 id;
@@ -27,6 +46,19 @@ public:
     static_assert(sizeof(InParameter) == 0x70,
                   "SplitterDestinationData::InParameter has the wrong size!");
 
+    struct InParameterVersion2b {
+        /* 0x00 */ u32 magic; // 'SNDD'
+        /* 0x04 */ s32 id;
+        /* 0x08 */ std::array<f32, MaxMixBuffers> mix_volumes;
+        /* 0x68 */ u32 mix_id;
+        /* 0x6C */ std::array<SplitterDestinationData::BiquadFilterParameter2, MaxBiquadFilters> biquad_filters;
+        /* 0x9C */ bool in_use;
+        /* 0x9D */ bool reset_prev_volume;
+        /* 0x9E */ u8 reserved[10];
+    };
+    static_assert(sizeof(InParameterVersion2b) == 0xA8,
+                  "SplitterDestinationData::InParameterVersion2b has the wrong size!");
+
     SplitterDestinationData(s32 id);
 
     /**
@@ -116,6 +148,20 @@ public:
      */
     void SetNext(SplitterDestinationData* next);
 
+    /**
+     * Get biquad filter parameters for this destination (REV15+).
+     *
+     * @return Span of biquad filter parameters.
+     */
+    std::span<BiquadFilterParameter2> GetBiquadFilters();
+
+    /**
+     * Get const biquad filter parameters for this destination (REV15+).
+     *
+     * @return Const span of biquad filter parameters.
+     */
+    std::span<const BiquadFilterParameter2> GetBiquadFilters() const;
+
 private:
     /// Id of this destination
     const s32 id;
@@ -125,6 +171,8 @@ private:
     std::array<f32, MaxMixBuffers> mix_volumes{0.0f};
     /// Previous mix volumes
     std::array<f32, MaxMixBuffers> prev_mix_volumes{0.0f};
+    /// Biquad filter parameters (REV15+)
+    std::array<BiquadFilterParameter2, MaxBiquadFilters> biquad_filters{};
     /// Next destination in the mix chain
     SplitterDestinationData* next{};
     /// Is this destination in use?

src/audio_core/renderer/voice/voice_info.h

@@ -135,6 +135,17 @@ public:
     static_assert(sizeof(BiquadFilterParameter) == 0xC,
                   "VoiceInfo::BiquadFilterParameter has the wrong size!");
 
+    struct BiquadFilterParameter2 {
+        /* 0x00 */ bool enabled;
+        /* 0x01 */ u8 reserved1;
+        /* 0x02 */ u8 reserved2;
+        /* 0x03 */ u8 reserved3;
+        /* 0x04 */ std::array<f32, 3> numerator;   // b0, b1, b2
+        /* 0x10 */ std::array<f32, 2> denominator; // a1, a2 (a0 = 1)
+    };
+    static_assert(sizeof(BiquadFilterParameter2) == 0x18,
+                  "VoiceInfo::BiquadFilterParameter2 has the wrong size!");
+
     struct InParameter {
         /* 0x000 */ u32 id;
         /* 0x004 */ u32 node_id;
@@ -168,6 +179,43 @@ public:
     };
     static_assert(sizeof(InParameter) == 0x170, "VoiceInfo::InParameter has the wrong size!");
 
+    struct InParameter2 {
+        /* 0x000 */ u32 id;
+        /* 0x004 */ u32 node_id;
+        /* 0x008 */ bool is_new;
+        /* 0x009 */ bool in_use;
+        /* 0x00A */ PlayState play_state;
+        /* 0x00B */ SampleFormat sample_format;
+        /* 0x00C */ u32 sample_rate;
+        /* 0x010 */ s32 priority;
+        /* 0x014 */ s32 sort_order;
+        /* 0x018 */ u32 channel_count;
+        /* 0x01C */ f32 pitch;
+        /* 0x020 */ f32 volume;
+        /* 0x024 */ std::array<BiquadFilterParameter2, MaxBiquadFilters> biquads;
+        /* 0x054 */ u32 wave_buffer_count;
+        /* 0x058 */ u32 wave_buffer_index;
+        /* 0x05C */ u32 reserved1;
+        /* 0x060 */ CpuAddr src_data_address;
+        /* 0x068 */ u64 src_data_size;
+        /* 0x070 */ u32 mix_id;
+        /* 0x074 */ u32 splitter_id;
+        /* 0x078 */ std::array<WaveBufferInternal, MaxWaveBuffers> wave_buffer_internal;
+        /* 0x158 */ std::array<s32, MaxChannels> channel_resource_ids;
+        /* 0x170 */ bool clear_voice_drop;
+        /* 0x171 */ u8 flush_buffer_count;
+        /* 0x172 */ u16 reserved2;
+        /* 0x174 */ Flags flags;
+        /* 0x175 */ u8 reserved3;
+        /* 0x176 */ SrcQuality src_quality;
+        /* 0x177 */ u8 reserved4;
+        /* 0x178 */ u32 external_context;
+        /* 0x17C */ u32 external_context_size;
+        /* 0x180 */ u32 reserved5;
+        /* 0x184 */ u32 reserved6;
+    };
+    static_assert(sizeof(InParameter2) == 0x188, "VoiceInfo::InParameter2 has the wrong size!");
+
     struct OutStatus {
         /* 0x00 */ u64 played_sample_count;
         /* 0x08 */ u32 wave_buffers_consumed;
@@ -349,6 +397,10 @@ public:
     f32 prev_volume{};
     /// Biquad filters for generating filter commands on this voice
     std::array<BiquadFilterParameter, MaxBiquadFilters> biquads{};
+    /// Float biquad filters for REV15+ (native float coefficients)
+    std::array<BiquadFilterParameter2, MaxBiquadFilters> biquads_float{};
+    /// Use float biquad coefficients (REV15+)
+    bool use_float_biquads{};
     /// Number of active wavebuffers
     u32 wave_buffer_count{};
     /// Current playing wavebuffer index

src/hid_core/resources/ring_lifo.h

@@ -1,4 +1,5 @@
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
+// SPDX-FileCopyrightText: Copyright 2025 Citron Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
@@ -44,8 +45,7 @@ struct Lifo {
             buffer_count++;
         }
         buffer_tail = GetNextEntryIndex();
-        const auto& previous_entry = ReadPreviousEntry();
+        entries[buffer_tail].sampling_number = new_state.sampling_number << 1;
-        entries[buffer_tail].sampling_number = previous_entry.sampling_number + 1;
         entries[buffer_tail].state = new_state;
     }
 };