6 files changed, 1106 insertions, 0 deletions

diff --git a/src/audio_core/renderer/voice/voice_channel_resource.h b/src/audio_core/renderer/voice/voice_channel_resource.h
new file mode 100644
index 000000000..26ab4ccce
--- /dev/null
+++ b/src/audio_core/renderer/voice/voice_channel_resource.h
@@ -0,0 +1,38 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+
+#include "audio_core/common/common.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+/**
+ * Represents one channel for mixing a voice.
+ */
+class VoiceChannelResource {
+public:
+    struct InParameter {
+        /* 0x00 */ u32 id;
+        /* 0x04 */ std::array<f32, MaxMixBuffers> mix_volumes;
+        /* 0x64 */ bool in_use;
+        /* 0x65 */ char unk65[0xB];
+    };
+    static_assert(sizeof(InParameter) == 0x70,
+                  "VoiceChannelResource::InParameter has the wrong size!");
+
+    explicit VoiceChannelResource(u32 id_) : id{id_} {}
+
+    /// Current volume for each mix buffer
+    std::array<f32, MaxMixBuffers> mix_volumes{};
+    /// Previous volume for each mix buffer
+    std::array<f32, MaxMixBuffers> prev_mix_volumes{};
+    /// Id of this resource
+    const u32 id;
+    /// Is this resource in use?
+    bool in_use{};
+};
+
+} // namespace AudioCore::AudioRenderer
diff --git a/src/audio_core/renderer/voice/voice_context.cpp b/src/audio_core/renderer/voice/voice_context.cpp
new file mode 100644
index 000000000..eafb51b01
--- /dev/null
+++ b/src/audio_core/renderer/voice/voice_context.cpp
@@ -0,0 +1,86 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <ranges>
+
+#include "audio_core/renderer/voice/voice_context.h"
+
+namespace AudioCore::AudioRenderer {
+
+VoiceState& VoiceContext::GetDspSharedState(const u32 index) {
+    if (index >= dsp_states.size()) {
+        LOG_ERROR(Service_Audio, "Invalid voice dsp state index {:04X}", index);
+    }
+    return dsp_states[index];
+}
+
+VoiceChannelResource& VoiceContext::GetChannelResource(const u32 index) {
+    if (index >= channel_resources.size()) {
+        LOG_ERROR(Service_Audio, "Invalid voice channel resource index {:04X}", index);
+    }
+    return channel_resources[index];
+}
+
+void VoiceContext::Initialize(std::span<VoiceInfo*> sorted_voice_infos_,
+                              std::span<VoiceInfo> voice_infos_,
+                              std::span<VoiceChannelResource> voice_channel_resources_,
+                              std::span<VoiceState> cpu_states_, std::span<VoiceState> dsp_states_,
+                              const u32 voice_count_) {
+    sorted_voice_info = sorted_voice_infos_;
+    voices = voice_infos_;
+    channel_resources = voice_channel_resources_;
+    cpu_states = cpu_states_;
+    dsp_states = dsp_states_;
+    voice_count = voice_count_;
+    active_count = 0;
+}
+
+VoiceInfo* VoiceContext::GetSortedInfo(const u32 index) {
+    if (index >= sorted_voice_info.size()) {
+        LOG_ERROR(Service_Audio, "Invalid voice sorted info index {:04X}", index);
+    }
+    return sorted_voice_info[index];
+}
+
+VoiceInfo& VoiceContext::GetInfo(const u32 index) {
+    if (index >= voices.size()) {
+        LOG_ERROR(Service_Audio, "Invalid voice info index {:04X}", index);
+    }
+    return voices[index];
+}
+
+VoiceState& VoiceContext::GetState(const u32 index) {
+    if (index >= cpu_states.size()) {
+        LOG_ERROR(Service_Audio, "Invalid voice cpu state index {:04X}", index);
+    }
+    return cpu_states[index];
+}
+
+u32 VoiceContext::GetCount() const {
+    return voice_count;
+}
+
+u32 VoiceContext::GetActiveCount() const {
+    return active_count;
+}
+
+void VoiceContext::SetActiveCount(const u32 active_count_) {
+    active_count = active_count_;
+}
+
+void VoiceContext::SortInfo() {
+    for (u32 i = 0; i < voice_count; i++) {
+        sorted_voice_info[i] = &voices[i];
+    }
+
+    std::ranges::sort(sorted_voice_info, [](const VoiceInfo* a, const VoiceInfo* b) {
+        return a->priority != b->priority ? a->priority < b->priority
+                                          : a->sort_order < b->sort_order;
+    });
+}
+
+void VoiceContext::UpdateStateByDspShared() {
+    std::memcpy(cpu_states.data(), dsp_states.data(), voice_count * sizeof(VoiceState));
+}
+
+} // namespace AudioCore::AudioRenderer
diff --git a/src/audio_core/renderer/voice/voice_context.h b/src/audio_core/renderer/voice/voice_context.h
new file mode 100644
index 000000000..43b677154
--- /dev/null
+++ b/src/audio_core/renderer/voice/voice_context.h
@@ -0,0 +1,126 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <span>
+
+#include "audio_core/renderer/voice/voice_channel_resource.h"
+#include "audio_core/renderer/voice/voice_info.h"
+#include "audio_core/renderer/voice/voice_state.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+/**
+ * Contains all voices, with utility functions for managing them.
+ */
+class VoiceContext {
+public:
+    /**
+     * Get the AudioRenderer state for a given index
+     *
+     * @param index - State index to get.
+     * @return The requested voice state.
+     */
+    VoiceState& GetDspSharedState(u32 index);
+
+    /**
+     * Get the channel resource for a given index
+     *
+     * @param index - Resource index to get.
+     * @return The requested voice resource.
+     */
+    VoiceChannelResource& GetChannelResource(u32 index);
+
+    /**
+     * Initialize the voice context.
+     *
+     * @param sorted_voice_infos      - Workbuffer for the sorted voices.
+     * @param voice_infos             - Workbuffer for the voices.
+     * @param voice_channel_resources - Workbuffer for the voice channel resources.
+     * @param cpu_states              - Workbuffer for the host-side voice states.
+     * @param dsp_states              - Workbuffer for the AudioRenderer-side voice states.
+     * @param voice_count             - The number of voices in each workbuffer.
+     */
+    void Initialize(std::span<VoiceInfo*> sorted_voice_infos, std::span<VoiceInfo> voice_infos,
+                    std::span<VoiceChannelResource> voice_channel_resources,
+                    std::span<VoiceState> cpu_states, std::span<VoiceState> dsp_states,
+                    u32 voice_count);
+
+    /**
+     * Get a sorted voice with the given index.
+     *
+     * @param index - The sorted voice index to get.
+     * @return The sorted voice.
+     */
+    VoiceInfo* GetSortedInfo(u32 index);
+
+    /**
+     * Get a voice with the given index.
+     *
+     * @param index - The voice index to get.
+     * @return The voice.
+     */
+    VoiceInfo& GetInfo(u32 index);
+
+    /**
+     * Get a host voice state with the given index.
+     *
+     * @param index - The host voice state index to get.
+     * @return The voice state.
+     */
+    VoiceState& GetState(u32 index);
+
+    /**
+     * Get the maximum number of voices.
+     * Not all voices in the buffers may be in use, see GetActiveCount.
+     *
+     * @return The maximum number of voices.
+     */
+    u32 GetCount() const;
+
+    /**
+     * Get the number of active voices.
+     * Can be less than or equal to the maximum number of voices.
+     *
+     * @return The number of active voices.
+     */
+    u32 GetActiveCount() const;
+
+    /**
+     * Set the number of active voices.
+     * Can be less than or equal to the maximum number of voices.
+     *
+     * @param active_count - The new number of active voices.
+     */
+    void SetActiveCount(u32 active_count);
+
+    /**
+     * Sort all voices. Results are available via GetSortedInfo.
+     * Voices are sorted descendingly, according to priority, and then sort order.
+     */
+    void SortInfo();
+
+    /**
+     * Update all voice states, copying AudioRenderer-side states to host-side states.
+     */
+    void UpdateStateByDspShared();
+
+private:
+    /// Sorted voices
+    std::span<VoiceInfo*> sorted_voice_info{};
+    /// Voices
+    std::span<VoiceInfo> voices{};
+    /// Channel resources
+    std::span<VoiceChannelResource> channel_resources{};
+    /// Host-side voice states
+    std::span<VoiceState> cpu_states{};
+    /// AudioRenderer-side voice states
+    std::span<VoiceState> dsp_states{};
+    /// Maximum number of voices
+    u32 voice_count{};
+    /// Number of active voices
+    u32 active_count{};
+};
+
+} // namespace AudioCore::AudioRenderer
diff --git a/src/audio_core/renderer/voice/voice_info.cpp b/src/audio_core/renderer/voice/voice_info.cpp
new file mode 100644
index 000000000..1849eeb57
--- /dev/null
+++ b/src/audio_core/renderer/voice/voice_info.cpp
@@ -0,0 +1,408 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "audio_core/renderer/memory/pool_mapper.h"
+#include "audio_core/renderer/voice/voice_context.h"
+#include "audio_core/renderer/voice/voice_info.h"
+#include "audio_core/renderer/voice/voice_state.h"
+
+namespace AudioCore::AudioRenderer {
+
+VoiceInfo::VoiceInfo() {
+    Initialize();
+}
+
+void VoiceInfo::Initialize() {
+    in_use = false;
+    is_new = false;
+    id = 0;
+    node_id = 0;
+    current_play_state = ServerPlayState::Stopped;
+    src_quality = SrcQuality::Medium;
+    priority = LowestVoicePriority;
+    sample_format = SampleFormat::Invalid;
+    sample_rate = 0;
+    channel_count = 0;
+    wave_buffer_count = 0;
+    wave_buffer_index = 0;
+    pitch = 0.0f;
+    volume = 0.0f;
+    prev_volume = 0.0f;
+    mix_id = UnusedMixId;
+    splitter_id = UnusedSplitterId;
+    biquads = {};
+    biquad_initialized = {};
+    voice_dropped = false;
+    data_unmapped = false;
+    buffer_unmapped = false;
+    flush_buffer_count = 0;
+
+    data_address.Setup(0, 0);
+    for (auto& wavebuffer : wavebuffers) {
+        wavebuffer.Initialize();
+    }
+}
+
+bool VoiceInfo::ShouldUpdateParameters(const InParameter& params) const {
+    return data_address.GetCpuAddr() != params.src_data_address ||
+           data_address.GetSize() != params.src_data_size || data_unmapped;
+}
+
+void VoiceInfo::UpdateParameters(BehaviorInfo::ErrorInfo& error_info, const InParameter& params,
+                                 const PoolMapper& pool_mapper, const BehaviorInfo& behavior) {
+    in_use = params.in_use;
+    id = params.id;
+    node_id = params.node_id;
+    UpdatePlayState(params.play_state);
+    UpdateSrcQuality(params.src_quality);
+    priority = params.priority;
+    sort_order = params.sort_order;
+    sample_rate = params.sample_rate;
+    sample_format = params.sample_format;
+    channel_count = static_cast<s8>(params.channel_count);
+    pitch = params.pitch;
+    volume = params.volume;
+    biquads = params.biquads;
+    wave_buffer_count = params.wave_buffer_count;
+    wave_buffer_index = params.wave_buffer_index;
+
+    if (behavior.IsFlushVoiceWaveBuffersSupported()) {
+        flush_buffer_count += params.flush_buffer_count;
+    }
+
+    mix_id = params.mix_id;
+
+    if (behavior.IsSplitterSupported()) {
+        splitter_id = params.splitter_id;
+    } else {
+        splitter_id = UnusedSplitterId;
+    }
+
+    channel_resource_ids = params.channel_resource_ids;
+
+    flags &= u16(~0b11);
+    if (behavior.IsVoicePlayedSampleCountResetAtLoopPointSupported()) {
+        flags |= u16(params.flags.IsVoicePlayedSampleCountResetAtLoopPointSupported);
+    }
+
+    if (behavior.IsVoicePitchAndSrcSkippedSupported()) {
+        flags |= u16(params.flags.IsVoicePitchAndSrcSkippedSupported);
+    }
+
+    if (params.clear_voice_drop) {
+        voice_dropped = false;
+    }
+
+    if (ShouldUpdateParameters(params)) {
+        data_unmapped = !pool_mapper.TryAttachBuffer(error_info, data_address,
+                                                     params.src_data_address, params.src_data_size);
+    } else {
+        error_info.error_code = ResultSuccess;
+        error_info.address = CpuAddr(0);
+    }
+}
+
+void VoiceInfo::UpdatePlayState(const PlayState state) {
+    last_play_state = current_play_state;
+
+    switch (state) {
+    case PlayState::Started:
+        current_play_state = ServerPlayState::Started;
+        break;
+    case PlayState::Stopped:
+        if (current_play_state != ServerPlayState::Stopped) {
+            current_play_state = ServerPlayState::RequestStop;
+        }
+        break;
+    case PlayState::Paused:
+        current_play_state = ServerPlayState::Paused;
+        break;
+    default:
+        LOG_ERROR(Service_Audio, "Invalid input play state {}", static_cast<u32>(state));
+        break;
+    }
+}
+
+void VoiceInfo::UpdateSrcQuality(const SrcQuality quality) {
+    switch (quality) {
+    case SrcQuality::Medium:
+        src_quality = quality;
+        break;
+    case SrcQuality::High:
+        src_quality = quality;
+        break;
+    case SrcQuality::Low:
+        src_quality = quality;
+        break;
+    default:
+        LOG_ERROR(Service_Audio, "Invalid input src quality {}", static_cast<u32>(quality));
+        break;
+    }
+}
+
+void VoiceInfo::UpdateWaveBuffers(std::span<std::array<BehaviorInfo::ErrorInfo, 2>> error_infos,
+                                  [[maybe_unused]] u32 error_count, const InParameter& params,
+                                  std::span<VoiceState*> voice_states,
+                                  const PoolMapper& pool_mapper, const BehaviorInfo& behavior) {
+    if (params.is_new) {
+        for (size_t i = 0; i < wavebuffers.size(); i++) {
+            wavebuffers[i].Initialize();
+        }
+
+        for (s8 channel = 0; channel < static_cast<s8>(params.channel_count); channel++) {
+            voice_states[channel]->wave_buffer_valid.fill(false);
+        }
+    }
+
+    for (u32 i = 0; i < MaxWaveBuffers; i++) {
+        UpdateWaveBuffer(error_infos[i], wavebuffers[i], params.wave_buffer_internal[i],
+                         params.sample_format, voice_states[0]->wave_buffer_valid[i], pool_mapper,
+                         behavior);
+    }
+}
+
+void VoiceInfo::UpdateWaveBuffer(std::span<BehaviorInfo::ErrorInfo> error_info,
+                                 WaveBuffer& wave_buffer,
+                                 const WaveBufferInternal& wave_buffer_internal,
+                                 const SampleFormat sample_format_, const bool valid,
+                                 const PoolMapper& pool_mapper, const BehaviorInfo& behavior) {
+    if (!valid && wave_buffer.sent_to_DSP && wave_buffer.buffer_address.GetCpuAddr() != 0) {
+        pool_mapper.ForceUnmapPointer(wave_buffer.buffer_address);
+        wave_buffer.buffer_address.Setup(0, 0);
+    }
+
+    if (!ShouldUpdateWaveBuffer(wave_buffer_internal)) {
+        return;
+    }
+
+    switch (sample_format_) {
+    case SampleFormat::PcmInt16: {
+        constexpr auto byte_size{GetSampleFormatByteSize(SampleFormat::PcmInt16)};
+        if (wave_buffer_internal.start_offset * byte_size > wave_buffer_internal.size ||
+            wave_buffer_internal.end_offset * byte_size > wave_buffer_internal.size) {
+            LOG_ERROR(Service_Audio, "Invalid PCM16 start/end wavebuffer sizes!");
+            error_info[0].error_code = Service::Audio::ERR_INVALID_UPDATE_DATA;
+            error_info[0].address = wave_buffer_internal.address;
+            return;
+        }
+    } break;
+
+    case SampleFormat::PcmFloat: {
+        constexpr auto byte_size{GetSampleFormatByteSize(SampleFormat::PcmFloat)};
+        if (wave_buffer_internal.start_offset * byte_size > wave_buffer_internal.size ||
+            wave_buffer_internal.end_offset * byte_size > wave_buffer_internal.size) {
+            LOG_ERROR(Service_Audio, "Invalid PCMFloat start/end wavebuffer sizes!");
+            error_info[0].error_code = Service::Audio::ERR_INVALID_UPDATE_DATA;
+            error_info[0].address = wave_buffer_internal.address;
+            return;
+        }
+    } break;
+
+    case SampleFormat::Adpcm: {
+        const auto start_frame{wave_buffer_internal.start_offset / 14};
+        auto start_extra{wave_buffer_internal.start_offset % 14 == 0
+                             ? 0
+                             : (wave_buffer_internal.start_offset % 14) / 2 + 1 +
+                                   ((wave_buffer_internal.start_offset % 14) % 2)};
+        const auto start{start_frame * 8 + start_extra};
+
+        const auto end_frame{wave_buffer_internal.end_offset / 14};
+        const auto end_extra{wave_buffer_internal.end_offset % 14 == 0
+                                 ? 0
+                                 : (wave_buffer_internal.end_offset % 14) / 2 + 1 +
+                                       ((wave_buffer_internal.end_offset % 14) % 2)};
+        const auto end{end_frame * 8 + end_extra};
+
+        if (start > static_cast<s64>(wave_buffer_internal.size) ||
+            end > static_cast<s64>(wave_buffer_internal.size)) {
+            LOG_ERROR(Service_Audio, "Invalid ADPCM start/end wavebuffer sizes!");
+            error_info[0].error_code = Service::Audio::ERR_INVALID_UPDATE_DATA;
+            error_info[0].address = wave_buffer_internal.address;
+            return;
+        }
+    } break;
+
+    default:
+        break;
+    }
+
+    if (wave_buffer_internal.start_offset < 0 || wave_buffer_internal.end_offset < 0) {
+        LOG_ERROR(Service_Audio, "Invalid input start/end wavebuffer sizes!");
+        error_info[0].error_code = Service::Audio::ERR_INVALID_UPDATE_DATA;
+        error_info[0].address = wave_buffer_internal.address;
+        return;
+    }
+
+    wave_buffer.start_offset = wave_buffer_internal.start_offset;
+    wave_buffer.end_offset = wave_buffer_internal.end_offset;
+    wave_buffer.loop = wave_buffer_internal.loop;
+    wave_buffer.stream_ended = wave_buffer_internal.stream_ended;
+    wave_buffer.sent_to_DSP = false;
+    wave_buffer.loop_start_offset = wave_buffer_internal.loop_start;
+    wave_buffer.loop_end_offset = wave_buffer_internal.loop_end;
+    wave_buffer.loop_count = wave_buffer_internal.loop_count;
+
+    buffer_unmapped =
+        !pool_mapper.TryAttachBuffer(error_info[0], wave_buffer.buffer_address,
+                                     wave_buffer_internal.address, wave_buffer_internal.size);
+
+    if (sample_format_ == SampleFormat::Adpcm && behavior.IsAdpcmLoopContextBugFixed() &&
+        wave_buffer_internal.context_address != 0) {
+        buffer_unmapped = !pool_mapper.TryAttachBuffer(error_info[1], wave_buffer.context_address,
+                                                       wave_buffer_internal.context_address,
+                                                       wave_buffer_internal.context_size) ||
+                          data_unmapped;
+    } else {
+        wave_buffer.context_address.Setup(0, 0);
+    }
+}
+
+bool VoiceInfo::ShouldUpdateWaveBuffer(const WaveBufferInternal& wave_buffer_internal) const {
+    return !wave_buffer_internal.sent_to_DSP || buffer_unmapped;
+}
+
+void VoiceInfo::WriteOutStatus(OutStatus& out_status, const InParameter& params,
+                               std::span<VoiceState*> voice_states) {
+    if (params.is_new) {
+        is_new = true;
+    }
+
+    if (params.is_new || is_new) {
+        out_status.played_sample_count = 0;
+        out_status.wave_buffers_consumed = 0;
+        out_status.voice_dropped = false;
+    } else {
+        out_status.played_sample_count = voice_states[0]->played_sample_count;
+        out_status.wave_buffers_consumed = voice_states[0]->wave_buffers_consumed;
+        out_status.voice_dropped = voice_dropped;
+    }
+}
+
+bool VoiceInfo::ShouldSkip() const {
+    return !in_use || wave_buffer_count == 0 || data_unmapped || buffer_unmapped || voice_dropped;
+}
+
+bool VoiceInfo::HasAnyConnection() const {
+    return mix_id != UnusedMixId || splitter_id != UnusedSplitterId;
+}
+
+void VoiceInfo::FlushWaveBuffers(const u32 flush_count, std::span<VoiceState*> voice_states,
+                                 const s8 channel_count_) {
+    auto wave_index{wave_buffer_index};
+
+    for (size_t i = 0; i < flush_count; i++) {
+        wavebuffers[wave_index].sent_to_DSP = true;
+
+        for (s8 j = 0; j < channel_count_; j++) {
+            auto voice_state{voice_states[j]};
+            if (voice_state->wave_buffer_index == wave_index) {
+                voice_state->wave_buffer_index =
+                    (voice_state->wave_buffer_index + 1) % MaxWaveBuffers;
+                voice_state->wave_buffers_consumed++;
+            }
+            voice_state->wave_buffer_valid[wave_index] = false;
+        }
+
+        wave_index = (wave_index + 1) % MaxWaveBuffers;
+    }
+}
+
+bool VoiceInfo::UpdateParametersForCommandGeneration(std::span<VoiceState*> voice_states) {
+    if (flush_buffer_count > 0) {
+        FlushWaveBuffers(flush_buffer_count, voice_states, channel_count);
+        flush_buffer_count = 0;
+    }
+
+    switch (current_play_state) {
+    case ServerPlayState::Started:
+        for (u32 i = 0; i < MaxWaveBuffers; i++) {
+            if (!wavebuffers[i].sent_to_DSP) {
+                for (s8 channel = 0; channel < channel_count; channel++) {
+                    voice_states[channel]->wave_buffer_valid[i] = true;
+                }
+                wavebuffers[i].sent_to_DSP = true;
+            }
+        }
+
+        was_playing = false;
+
+        for (u32 i = 0; i < MaxWaveBuffers; i++) {
+            if (voice_states[0]->wave_buffer_valid[i]) {
+                return true;
+            }
+        }
+        break;
+
+    case ServerPlayState::Stopped:
+    case ServerPlayState::Paused:
+        for (auto& wavebuffer : wavebuffers) {
+            if (!wavebuffer.sent_to_DSP) {
+                wavebuffer.buffer_address.GetReference(true);
+                wavebuffer.context_address.GetReference(true);
+            }
+        }
+
+        if (sample_format == SampleFormat::Adpcm && data_address.GetCpuAddr() != 0) {
+            data_address.GetReference(true);
+        }
+
+        was_playing = last_play_state == ServerPlayState::Started;
+        break;
+
+    case ServerPlayState::RequestStop:
+        for (u32 i = 0; i < MaxWaveBuffers; i++) {
+            wavebuffers[i].sent_to_DSP = true;
+
+            for (s8 channel = 0; channel < channel_count; channel++) {
+                if (voice_states[channel]->wave_buffer_valid[i]) {
+                    voice_states[channel]->wave_buffer_index =
+                        (voice_states[channel]->wave_buffer_index + 1) % MaxWaveBuffers;
+                    voice_states[channel]->wave_buffers_consumed++;
+                }
+                voice_states[channel]->wave_buffer_valid[i] = false;
+            }
+        }
+
+        for (s8 channel = 0; channel < channel_count; channel++) {
+            voice_states[channel]->offset = 0;
+            voice_states[channel]->played_sample_count = 0;
+            voice_states[channel]->adpcm_context = {};
+            voice_states[channel]->sample_history.fill(0);
+            voice_states[channel]->fraction = 0;
+        }
+
+        current_play_state = ServerPlayState::Stopped;
+        was_playing = last_play_state == ServerPlayState::Started;
+        break;
+    }
+
+    return was_playing;
+}
+
+bool VoiceInfo::UpdateForCommandGeneration(VoiceContext& voice_context) {
+    std::array<VoiceState*, MaxChannels> voice_states{};
+
+    if (is_new) {
+        ResetResources(voice_context);
+        prev_volume = volume;
+        is_new = false;
+    }
+
+    for (s8 channel = 0; channel < channel_count; channel++) {
+        voice_states[channel] = &voice_context.GetDspSharedState(channel_resource_ids[channel]);
+    }
+
+    return UpdateParametersForCommandGeneration(voice_states);
+}
+
+void VoiceInfo::ResetResources(VoiceContext& voice_context) const {
+    for (s8 channel = 0; channel < channel_count; channel++) {
+        auto& state{voice_context.GetDspSharedState(channel_resource_ids[channel])};
+        state = {};
+
+        auto& channel_resource{voice_context.GetChannelResource(channel_resource_ids[channel])};
+        channel_resource.prev_mix_volumes = channel_resource.mix_volumes;
+    }
+}
+
+} // namespace AudioCore::AudioRenderer
diff --git a/src/audio_core/renderer/voice/voice_info.h b/src/audio_core/renderer/voice/voice_info.h
new file mode 100644
index 000000000..896723e0c
--- /dev/null
+++ b/src/audio_core/renderer/voice/voice_info.h
@@ -0,0 +1,378 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <bitset>
+
+#include "audio_core/common/common.h"
+#include "audio_core/common/wave_buffer.h"
+#include "audio_core/renderer/behavior/behavior_info.h"
+#include "audio_core/renderer/memory/address_info.h"
+#include "common/common_types.h"
+
+namespace AudioCore::AudioRenderer {
+class PoolMapper;
+class VoiceContext;
+struct VoiceState;
+
+/**
+ * Represents one voice. Voices are essentially noises, and they can be further mixed and have
+ * effects applied to them, but voices are the basis of all sounds.
+ */
+class VoiceInfo {
+public:
+    enum class ServerPlayState {
+        Started,
+        Stopped,
+        RequestStop,
+        Paused,
+    };
+
+    struct Flags {
+        u8 IsVoicePlayedSampleCountResetAtLoopPointSupported : 1;
+        u8 IsVoicePitchAndSrcSkippedSupported : 1;
+    };
+
+    /**
+     * A wavebuffer contains information on the data source buffers.
+     */
+    struct WaveBuffer {
+        void Copy(WaveBufferVersion1& other) {
+            other.buffer = buffer_address.GetReference(true);
+            other.buffer_size = buffer_address.GetSize();
+            other.start_offset = start_offset;
+            other.end_offset = end_offset;
+            other.loop = loop;
+            other.stream_ended = stream_ended;
+
+            if (context_address.GetCpuAddr()) {
+                other.context = context_address.GetReference(true);
+                other.context_size = context_address.GetSize();
+            } else {
+                other.context = CpuAddr(0);
+                other.context_size = 0;
+            }
+        }
+
+        void Copy(WaveBufferVersion2& other) {
+            other.buffer = buffer_address.GetReference(true);
+            other.buffer_size = buffer_address.GetSize();
+            other.start_offset = start_offset;
+            other.end_offset = end_offset;
+            other.loop_start_offset = loop_start_offset;
+            other.loop_end_offset = loop_end_offset;
+            other.loop = loop;
+            other.loop_count = loop_count;
+            other.stream_ended = stream_ended;
+
+            if (context_address.GetCpuAddr()) {
+                other.context = context_address.GetReference(true);
+                other.context_size = context_address.GetSize();
+            } else {
+                other.context = CpuAddr(0);
+                other.context_size = 0;
+            }
+        }
+
+        void Initialize() {
+            buffer_address.Setup(0, 0);
+            context_address.Setup(0, 0);
+            start_offset = 0;
+            end_offset = 0;
+            loop = false;
+            stream_ended = false;
+            sent_to_DSP = true;
+            loop_start_offset = 0;
+            loop_end_offset = 0;
+            loop_count = 0;
+        }
+        /// Game memory address of the wavebuffer data
+        AddressInfo buffer_address{0, 0};
+        /// Context for decoding, used for ADPCM
+        AddressInfo context_address{0, 0};
+        /// Starting offset for the wavebuffer
+        u32 start_offset{};
+        /// Ending offset the wavebuffer
+        u32 end_offset{};
+        /// Should this wavebuffer loop?
+        bool loop{};
+        /// Has this wavebuffer ended?
+        bool stream_ended{};
+        /// Has this wavebuffer been sent to the AudioRenderer?
+        bool sent_to_DSP{true};
+        /// Starting offset when looping, can differ from start_offset
+        u32 loop_start_offset{};
+        /// Ending offset when looping, can differ from end_offset
+        u32 loop_end_offset{};
+        /// Number of times to loop this wavebuffer
+        s32 loop_count{};
+    };
+
+    struct WaveBufferInternal {
+        /* 0x00 */ CpuAddr address;
+        /* 0x08 */ u64 size;
+        /* 0x10 */ s32 start_offset;
+        /* 0x14 */ s32 end_offset;
+        /* 0x18 */ bool loop;
+        /* 0x19 */ bool stream_ended;
+        /* 0x1A */ bool sent_to_DSP;
+        /* 0x1C */ s32 loop_count;
+        /* 0x20 */ CpuAddr context_address;
+        /* 0x28 */ u64 context_size;
+        /* 0x30 */ u32 loop_start;
+        /* 0x34 */ u32 loop_end;
+    };
+    static_assert(sizeof(WaveBufferInternal) == 0x38,
+                  "VoiceInfo::WaveBufferInternal has the wrong size!");
+
+    struct BiquadFilterParameter {
+        /* 0x00 */ bool enabled;
+        /* 0x02 */ std::array<s16, 3> b;
+        /* 0x08 */ std::array<s16, 2> a;
+    };
+    static_assert(sizeof(BiquadFilterParameter) == 0xC,
+                  "VoiceInfo::BiquadFilterParameter has the wrong size!");
+
+    struct InParameter {
+        /* 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<BiquadFilterParameter, MaxBiquadFilters> biquads;
+        /* 0x03C */ u32 wave_buffer_count;
+        /* 0x040 */ u16 wave_buffer_index;
+        /* 0x042 */ char unk042[0x6];
+        /* 0x048 */ CpuAddr src_data_address;
+        /* 0x050 */ u64 src_data_size;
+        /* 0x058 */ u32 mix_id;
+        /* 0x05C */ u32 splitter_id;
+        /* 0x060 */ std::array<WaveBufferInternal, MaxWaveBuffers> wave_buffer_internal;
+        /* 0x140 */ std::array<u32, MaxChannels> channel_resource_ids;
+        /* 0x158 */ bool clear_voice_drop;
+        /* 0x159 */ u8 flush_buffer_count;
+        /* 0x15A */ char unk15A[0x2];
+        /* 0x15C */ Flags flags;
+        /* 0x15D */ char unk15D[0x1];
+        /* 0x15E */ SrcQuality src_quality;
+        /* 0x15F */ char unk15F[0x11];
+    };
+    static_assert(sizeof(InParameter) == 0x170, "VoiceInfo::InParameter has the wrong size!");
+
+    struct OutStatus {
+        /* 0x00 */ u64 played_sample_count;
+        /* 0x08 */ u32 wave_buffers_consumed;
+        /* 0x0C */ bool voice_dropped;
+    };
+    static_assert(sizeof(OutStatus) == 0x10, "OutStatus::InParameter has the wrong size!");
+
+    VoiceInfo();
+
+    /**
+     * Initialize this voice.
+     */
+    void Initialize();
+
+    /**
+     * Does this voice ned an update?
+     *
+     * @param params - Input parametetrs to check matching.
+     * @return True if this voice needs an update, otherwise false.
+     */
+    bool ShouldUpdateParameters(const InParameter& params) const;
+
+    /**
+     * Update the parameters of this voice.
+     *
+     * @param error_info  - Output error code.
+     * @param params      - Input parametters to udpate from.
+     * @param pool_mapper - Used to map buffers.
+     * @param behavior   - behavior to check supported features.
+     */
+    void UpdateParameters(BehaviorInfo::ErrorInfo& error_info, const InParameter& params,
+                          const PoolMapper& pool_mapper, const BehaviorInfo& behavior);
+
+    /**
+     * Update the current play state.
+     *
+     * @param state - New play state for this voice.
+     */
+    void UpdatePlayState(PlayState state);
+
+    /**
+     * Update the current sample rate conversion quality.
+     *
+     * @param quality - New quality.
+     */
+    void UpdateSrcQuality(SrcQuality quality);
+
+    /**
+     * Update all wavebuffers.
+     *
+     * @param error_infos - Output 2D array of errors, 2 per wavebuffer.
+     * @param error_count - Number of errors provided. Unused.
+     * @param params - Input parametters to be used for the update.
+     * @param voice_states - The voice states for each channel in this voice to be updated.
+     * @param pool_mapper - Used to map the wavebuffers.
+     * @param behavior - Used to check for supported features.
+     */
+    void UpdateWaveBuffers(std::span<std::array<BehaviorInfo::ErrorInfo, 2>> error_infos,
+                           u32 error_count, const InParameter& params,
+                           std::span<VoiceState*> voice_states, const PoolMapper& pool_mapper,
+                           const BehaviorInfo& behavior);
+
+    /**
+     * Update a wavebuffer.
+     *
+     * @param error_infos          - Output array of errors.
+     * @param wave_buffer          - The wavebuffer to be updated.
+     * @param wave_buffer_internal - Input parametters to be used for the update.
+     * @param sample_format        - Sample format of the wavebuffer.
+     * @param valid                - Is this wavebuffer valid?
+     * @param pool_mapper          - Used to map the wavebuffers.
+     * @param behavior            - Used to check for supported features.
+     */
+    void UpdateWaveBuffer(std::span<BehaviorInfo::ErrorInfo> error_info, WaveBuffer& wave_buffer,
+                          const WaveBufferInternal& wave_buffer_internal,
+                          SampleFormat sample_format, bool valid, const PoolMapper& pool_mapper,
+                          const BehaviorInfo& behavior);
+
+    /**
+     * Check if the input wavebuffer needs an update.
+     *
+     * @param wave_buffer_internal - Input wavebuffer parameters to check.
+     * @return True if the given wavebuffer needs an update, otherwise false.
+     */
+    bool ShouldUpdateWaveBuffer(const WaveBufferInternal& wave_buffer_internal) const;
+
+    /**
+     * Write the number of played samples, number of consumed wavebuffers and if this voice was
+     * dropped, to the given out_status.
+     *
+     * @param out_status   - Output status to be written to.
+     * @param in_params    - Input parameters to check if the wavebuffer is new.
+     * @param voice_states - Current host voice states for this voice, source of the output.
+     */
+    void WriteOutStatus(OutStatus& out_status, const InParameter& in_params,
+                        std::span<VoiceState*> voice_states);
+
+    /**
+     * Check if this voice should be skipped for command generation.
+     * Checks various things such as usage state, whether data is mapped etc.
+     *
+     * @return True if this voice should not be generated, otherwise false.
+     */
+    bool ShouldSkip() const;
+
+    /**
+     * Check if this voice has any mixing connections.
+     *
+     * @return True if this voice participes in mixing, otherwise false.
+     */
+    bool HasAnyConnection() const;
+
+    /**
+     * Flush flush_count wavebuffers, marking them as consumed.
+     *
+     * @param flush_count   - Number of wavebuffers to flush.
+     * @param voice_states  - Voice states for these wavebuffers.
+     * @param channel_count - Number of active channels.
+     */
+    void FlushWaveBuffers(u32 flush_count, std::span<VoiceState*> voice_states, s8 channel_count);
+
+    /**
+     * Update this voice's parameters on command generation,
+     * updating voice states and flushing if needed.
+     *
+     * @param voice_states  - Voice states for these wavebuffers.
+     * @return True if this voice should be generated, otherwise false.
+     */
+    bool UpdateParametersForCommandGeneration(std::span<VoiceState*> voice_states);
+
+    /**
+     * Update this voice on command generation.
+     *
+     * @param voice_states  - Voice states for these wavebuffers.
+     * @return True if this voice should be generated, otherwise false.
+     */
+    bool UpdateForCommandGeneration(VoiceContext& voice_context);
+
+    /**
+     * Reset the AudioRenderer-side voice states, and the channel resources for this voice.
+     *
+     * @param voice_context - Context from which to get the resources.
+     */
+    void ResetResources(VoiceContext& voice_context) const;
+
+    /// Is this voice in use?
+    bool in_use{};
+    /// Is this voice new?
+    bool is_new{};
+    /// Was this voice last playing? Used for depopping
+    bool was_playing{};
+    /// Sample format of the wavebuffers in this voice
+    SampleFormat sample_format{};
+    /// Sample rate of the wavebuffers in this voice
+    u32 sample_rate{};
+    /// Number of channels in this voice
+    s8 channel_count{};
+    /// Id of this voice
+    u32 id{};
+    /// Node id of this voice
+    u32 node_id{};
+    /// Mix id this voice is mixed to
+    u32 mix_id{};
+    /// Play state of this voice
+    ServerPlayState current_play_state{ServerPlayState::Stopped};
+    /// Last play state of this voice
+    ServerPlayState last_play_state{ServerPlayState::Started};
+    /// Priority of this voice, lower is higher
+    s32 priority{};
+    /// Sort order of this voice, used when same priority
+    s32 sort_order{};
+    /// Pitch of this voice (for sample rate conversion)
+    f32 pitch{};
+    /// Current volume of this voice
+    f32 volume{};
+    /// Previous volume of this voice
+    f32 prev_volume{};
+    /// Biquad filters for generating filter commands on this voice
+    std::array<BiquadFilterParameter, MaxBiquadFilters> biquads{};
+    /// Number of active wavebuffers
+    u32 wave_buffer_count{};
+    /// Current playing wavebuffer index
+    u16 wave_buffer_index{};
+    /// Flags controlling decode behavior
+    u16 flags{};
+    /// Game memory for ADPCM coefficients
+    AddressInfo data_address{0, 0};
+    /// Wavebuffers
+    std::array<WaveBuffer, MaxWaveBuffers> wavebuffers{};
+    /// Channel resources for this voice
+    std::array<u32, MaxChannels> channel_resource_ids{};
+    /// Splitter id this voice is connected with
+    s32 splitter_id{UnusedSplitterId};
+    /// Sample rate conversion quality
+    SrcQuality src_quality{SrcQuality::Medium};
+    /// Was this voice dropped due to limited time?
+    bool voice_dropped{};
+    /// Is this voice's coefficient (data_address) unmapped?
+    bool data_unmapped{};
+    /// Is this voice's buffers (wavebuffer data and ADPCM context) unmapped?
+    bool buffer_unmapped{};
+    /// Initialisation state of the biquads
+    std::array<bool, MaxBiquadFilters> biquad_initialized{};
+    /// Number of wavebuffers to flush
+    u8 flush_buffer_count{};
+};
+
+} // namespace AudioCore::AudioRenderer
diff --git a/src/audio_core/renderer/voice/voice_state.h b/src/audio_core/renderer/voice/voice_state.h
new file mode 100644
index 000000000..d5497e2fb
--- /dev/null
+++ b/src/audio_core/renderer/voice/voice_state.h
@@ -0,0 +1,70 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+
+#include "audio_core/common/common.h"
+#include "common/common_types.h"
+#include "common/fixed_point.h"
+
+namespace AudioCore::AudioRenderer {
+/**
+ * Holds a state for a voice. One is kept host-side, and one is used by the AudioRenderer,
+ * host-side is updated on the next iteration.
+ */
+struct VoiceState {
+    /**
+     * State of the voice's biquad filter.
+     */
+    struct BiquadFilterState {
+        Common::FixedPoint<50, 14> s0;
+        Common::FixedPoint<50, 14> s1;
+        Common::FixedPoint<50, 14> s2;
+        Common::FixedPoint<50, 14> s3;
+    };
+
+    /**
+     * Context for ADPCM decoding.
+     */
+    struct AdpcmContext {
+        u16 header;
+        s16 yn0;
+        s16 yn1;
+    };
+
+    /// Number of samples played
+    u64 played_sample_count;
+    /// Current offset from the starting offset
+    u32 offset;
+    /// Currently active wavebuffer index
+    u32 wave_buffer_index;
+    /// Array of which wavebuffers are currently valid
+
+    std::array<bool, MaxWaveBuffers> wave_buffer_valid;
+    /// Number of wavebuffers consumed, given back to the game
+    u32 wave_buffers_consumed;
+    /// History of samples, used for rate conversion
+
+    std::array<s16, MaxWaveBuffers * 2> sample_history;
+    /// Current read fraction, used for resampling
+    Common::FixedPoint<49, 15> fraction;
+    /// Current adpcm context
+    AdpcmContext adpcm_context;
+    /// Current biquad states, used when filtering
+
+    std::array<std::array<BiquadFilterState, MaxBiquadFilters>, MaxBiquadFilters> biquad_states;
+    /// Previous samples
+    std::array<s32, MaxMixBuffers> previous_samples;
+    /// Unused
+    u32 external_context_size;
+    /// Unused
+    bool external_context_enabled;
+    /// Was this voice dropped?
+    bool voice_dropped;
+    /// Number of times the wavebuffer has looped
+    s32 loop_count;
+};
+
+} // namespace AudioCore::AudioRenderer