[bink] synced host movie playback to the guest audio clock

src/SharpEmu.Libs/Bink/Bink2MovieBridge.cs

@@ -133,10 +133,12 @@ internal static class Bink2MovieBridge
                     if (_playback.IsFinished)
                     {
                         var completedPath = _activePath;
+                        var progress = _playback.PlaybackProgress;
                         CloseActiveLocked();
                         Console.Error.WriteLine(
                             "[LOADER][INFO] Bink2 bridge completed: " +
-                            Path.GetFileName(completedPath));
+                            $"{Path.GetFileName(completedPath)} after " +
+                            $"{progress.Seconds:F2}s at frame {progress.FrameIndex}");
                         AttachNextQueuedMovieLocked();
                     }
                     return false;

src/SharpEmu.Libs/Bink/BinkFramePlayback.cs

@@ -2,6 +2,7 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 
 using System.Diagnostics;
+using SharpEmu.HLE.Host;
 
 namespace SharpEmu.Libs.Bink;
 
@@ -36,6 +37,8 @@ internal sealed class BinkFramePlayback : IDisposable
     private long _currentFrameIndex = -1;
     private long _nextDecodedFrameIndex;
     private long _playbackStartTimestamp;
+    private double _audioStartSeconds;
+    private long _lastSkewTraceTimestamp;
     private bool _playbackClockStarted;
     private bool _decoderCompleted;
     private bool _stopRequested;
@@ -83,6 +86,26 @@ internal sealed class BinkFramePlayback : IDisposable
         }
     }
 
+    /// <summary>
+    /// Wall-clock seconds since the first frame was presented, and the index of
+    /// the last frame shown. Playback is on its own time base when these agree
+    /// with the movie's frame rate.
+    /// </summary>
+    internal (double Seconds, long FrameIndex) PlaybackProgress
+    {
+        get
+        {
+            lock (_gate)
+            {
+                return (
+                    _playbackClockStarted
+                        ? Stopwatch.GetElapsedTime(_playbackStartTimestamp).TotalSeconds
+                        : 0,
+                    _currentFrameIndex);
+            }
+        }
+    }
+
     internal bool TryGetFrame(
         bool advanceClock,
         out byte[] pixels,
@@ -118,14 +141,13 @@ internal sealed class BinkFramePlayback : IDisposable
             if (advanceClock && !_playbackClockStarted)
             {
                 _playbackStartTimestamp = Stopwatch.GetTimestamp();
+                _audioStartSeconds = GuestAudioClock.PlayedSeconds;
                 _playbackClockStarted = true;
             }
 
-            var elapsedSeconds = _playbackClockStarted
-                ? Stopwatch.GetElapsedTime(_playbackStartTimestamp).TotalSeconds
-                : 0;
-            var targetFrameIndex = (long)Math.Floor(
-                elapsedSeconds * FramesPerSecondNumerator / FramesPerSecondDenominator);
+            var elapsedSeconds = CurrentPlaybackSecondsLocked();
+            TraceClockSkewLocked();
+            var targetFrameIndex = CurrentTargetFrameIndexLocked();
             DecodedFrame? replacement = null;
             while (_decodedFrames.Count > 0 &&
                    _decodedFrames.Peek().Index <= targetFrameIndex)
@@ -166,6 +188,86 @@ internal sealed class BinkFramePlayback : IDisposable
         }
     }
 
+    /// <summary>
+    /// Time base for playback. A host-decoded movie runs on whatever clock it is
+    /// given, but the audio that belongs to it comes from the guest, which does
+    /// not advance at wall-clock rate on a slow frame. Following the audio keeps
+    /// the two together; SHARPEMU_MOVIE_CLOCK=wall restores the old behaviour.
+    /// </summary>
+    private static readonly bool _followGuestAudioClock = !string.Equals(
+        Environment.GetEnvironmentVariable("SHARPEMU_MOVIE_CLOCK"),
+        "wall",
+        StringComparison.OrdinalIgnoreCase);
+
+    /// <summary>
+    /// Seconds of playback elapsed on the movie's time base. Falls back to wall
+    /// clock whenever guest audio is not flowing: a movie whose audio never
+    /// starts — or stops early — must still finish rather than hang on a clock
+    /// that will never advance again.
+    /// </summary>
+    private double CurrentPlaybackSecondsLocked()
+    {
+        if (!_playbackClockStarted)
+        {
+            return 0;
+        }
+
+        var wallSeconds = Stopwatch.GetElapsedTime(_playbackStartTimestamp).TotalSeconds;
+        if (!_followGuestAudioClock || !GuestAudioClock.IsRunning)
+        {
+            return wallSeconds;
+        }
+
+        return Math.Clamp(GuestAudioClock.PlayedSeconds - _audioStartSeconds, 0, wallSeconds);
+    }
+
+    private static readonly bool _traceClockSkew = string.Equals(
+        Environment.GetEnvironmentVariable("SHARPEMU_LOG_MOVIE_SYNC"),
+        "1",
+        StringComparison.Ordinal);
+
+    /// <summary>
+    /// Logs how far the movie's wall clock has drifted from the guest audio the
+    /// movie is supposed to be in step with. A skew that is flat across playback
+    /// is a late audio start; one that grows is a rate mismatch, and the two need
+    /// different fixes. Caller holds <see cref="_gate"/>.
+    /// </summary>
+    private void TraceClockSkewLocked()
+    {
+        if (!_traceClockSkew || !_playbackClockStarted)
+        {
+            return;
+        }
+
+        var now = Stopwatch.GetTimestamp();
+        if (_lastSkewTraceTimestamp != 0 &&
+            Stopwatch.GetElapsedTime(_lastSkewTraceTimestamp) < TimeSpan.FromSeconds(1))
+        {
+            return;
+        }
+
+        _lastSkewTraceTimestamp = now;
+        var wallSeconds = Stopwatch.GetElapsedTime(_playbackStartTimestamp).TotalSeconds;
+        var audioSeconds = GuestAudioClock.PlayedSeconds - _audioStartSeconds;
+        Console.Error.WriteLine(
+            $"[PERF][MOVIE] wall_s={wallSeconds:F2} audio_s={audioSeconds:F2} " +
+            $"playback_s={CurrentPlaybackSecondsLocked():F2} " +
+            $"skew_s={wallSeconds - audioSeconds:F2} frame={_currentFrameIndex} " +
+            $"audio_running={GuestAudioClock.IsRunning}");
+    }
+
+    /// <summary>
+    /// The frame the movie's own time base says should be on screen right now.
+    /// Returns -1 until the first frame is presented, so the queue prefills
+    /// instead of instantly declaring everything late.
+    /// </summary>
+    private long CurrentTargetFrameIndexLocked() =>
+        _playbackClockStarted
+            ? (long)Math.Floor(
+                CurrentPlaybackSecondsLocked() *
+                FramesPerSecondNumerator / FramesPerSecondDenominator)
+            : -1;
+
     private void DecodeLoop()
     {
         try
@@ -199,8 +301,28 @@ internal sealed class BinkFramePlayback : IDisposable
 
                 lock (_gate)
                 {
-                    _decodedFrames.Enqueue(new DecodedFrame(
-                        _nextDecodedFrameIndex++, destination));
+                    var frameIndex = _nextDecodedFrameIndex++;
+
+                    // Frames are pulled once per guest flip, so a title running
+                    // well under the movie's frame rate cannot drain a queue
+                    // this shallow fast enough and the movie stretches past its
+                    // real duration — audio finishes while the last picture sits
+                    // on screen and the next movie starts late. Once the clock
+                    // has passed a queued frame it can never be shown, so retire
+                    // it in favour of this newer one. Only superseded frames are
+                    // dropped, never the newest, so a decoder that cannot keep
+                    // up still advances the picture instead of freezing it.
+                    var targetFrameIndex = CurrentTargetFrameIndexLocked();
+                    if (frameIndex <= targetFrameIndex)
+                    {
+                        while (_decodedFrames.Count > 0 &&
+                               _decodedFrames.Peek().Index <= targetFrameIndex)
+                        {
+                            _freeBuffers.Enqueue(_decodedFrames.Dequeue().Pixels);
+                        }
+                    }
+
+                    _decodedFrames.Enqueue(new DecodedFrame(frameIndex, destination));
                     Monitor.PulseAll(_gate);
                 }
             }

src/SharpEmu.Libs/Bink/FfmpegNativeBinkFrameSource.cs

@@ -1,7 +1,9 @@
 // Copyright (C) 2026 SharpEmu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
+using System.Buffers;
 using FFmpeg.AutoGen;
+using SharpEmu.HLE.Host;
 
 namespace SharpEmu.Libs.Bink;
 
@@ -13,13 +15,29 @@ namespace SharpEmu.Libs.Bink;
 /// </summary>
 internal sealed unsafe class FfmpegNativeBinkFrameSource : IBinkFrameDecoder
 {
+    private const int OutputAudioChannels = 2;
+    private const int OutputAudioBytesPerSample = sizeof(short);
+
+    private readonly object _decodeGate = new();
     private AVFormatContext* _formatContext;
     private AVCodecContext* _codecContext;
+    private AVCodecContext* _audioCodecContext;
     private SwsContext* _swsContext;
+    private SwrContext* _swrContext;
     private AVFrame* _frame;
+    private AVFrame* _audioFrame;
     private AVPacket* _packet;
+    private IHostAudioStream? _audioStream;
     private readonly int _videoStreamIndex;
+    private readonly int _audioStreamIndex;
+    private readonly int _audioOutputSampleRate;
+    private AVChannelLayout _swrInputLayout;
+    private AVSampleFormat _swrInputFormat = AVSampleFormat.AV_SAMPLE_FMT_NONE;
+    private int _swrInputSampleRate;
+    private bool _swrInputLayoutValid;
     private bool _draining;
+    private bool _audioDraining;
+    private bool _audioFailed;
     private int _disposed;
 
     public uint Width { get; }
@@ -34,6 +52,10 @@ internal sealed unsafe class FfmpegNativeBinkFrameSource : IBinkFrameDecoder
         AVFormatContext* formatContext,
         AVCodecContext* codecContext,
         int videoStreamIndex,
+        AVCodecContext* audioCodecContext,
+        int audioStreamIndex,
+        IHostAudioStream? audioStream,
+        int audioOutputSampleRate,
         uint width,
         uint height,
         uint framesPerSecondNumerator,
@@ -42,11 +64,16 @@ internal sealed unsafe class FfmpegNativeBinkFrameSource : IBinkFrameDecoder
         _formatContext = formatContext;
         _codecContext = codecContext;
         _videoStreamIndex = videoStreamIndex;
+        _audioCodecContext = audioCodecContext;
+        _audioStreamIndex = audioStreamIndex;
+        _audioStream = audioStream;
+        _audioOutputSampleRate = audioOutputSampleRate;
         Width = width;
         Height = height;
         FramesPerSecondNumerator = framesPerSecondNumerator;
         FramesPerSecondDenominator = framesPerSecondDenominator;
         _frame = ffmpeg.av_frame_alloc();
+        _audioFrame = audioCodecContext is null ? null : ffmpeg.av_frame_alloc();
         _packet = ffmpeg.av_packet_alloc();
     }
 
@@ -93,6 +120,8 @@ internal sealed unsafe class FfmpegNativeBinkFrameSource : IBinkFrameDecoder
 
         AVFormatContext* formatContext = null;
         AVCodecContext* codecContext = null;
+        AVCodecContext* audioCodecContext = null;
+        IHostAudioStream? audioStream = null;
         try
         {
             if (ffmpeg.avformat_open_input(&formatContext, path, null, null) < 0)
@@ -151,6 +180,28 @@ internal sealed unsafe class FfmpegNativeBinkFrameSource : IBinkFrameDecoder
                 frameRate = new AVRational { num = 30, den = 1 };
             }
 
+            var audioStreamIndex = TryOpenAudioDecoder(
+                formatContext,
+                out audioCodecContext,
+                out var audioOutputSampleRate);
+            if (audioStreamIndex >= 0 && audioCodecContext is not null)
+            {
+                try
+                {
+                    audioStream = HostPlatform.Current.Audio.OpenStereoPcm16Stream(
+                        checked((uint)audioOutputSampleRate));
+                }
+                catch (Exception exception) when (exception is InvalidOperationException or
+                                                     ArgumentOutOfRangeException)
+                {
+                    Console.Error.WriteLine(
+                        $"[LOADER][WARN] Bink audio output unavailable: {exception.Message}");
+                    ffmpeg.avcodec_free_context(&audioCodecContext);
+                    audioStreamIndex = -1;
+                    audioOutputSampleRate = 0;
+                }
+            }
+
             var outputWidth = (uint)codecContext->width;
             var outputHeight = (uint)codecContext->height;
             if (maximumWidth > 0 && maximumHeight > 0 &&
@@ -175,12 +226,18 @@ internal sealed unsafe class FfmpegNativeBinkFrameSource : IBinkFrameDecoder
                 formatContext,
                 codecContext,
                 videoStreamIndex,
+                audioCodecContext,
+                audioStreamIndex,
+                audioStream,
+                audioOutputSampleRate,
                 outputWidth,
                 outputHeight,
                 (uint)frameRate.num,
                 (uint)frameRate.den);
             formatContext = null;
             codecContext = null;
+            audioCodecContext = null;
+            audioStream = null;
             return true;
         }
         catch (DllNotFoundException)
@@ -194,6 +251,13 @@ internal sealed unsafe class FfmpegNativeBinkFrameSource : IBinkFrameDecoder
                 ffmpeg.avcodec_free_context(&codecContext);
             }
 
+            if (audioCodecContext is not null)
+            {
+                ffmpeg.avcodec_free_context(&audioCodecContext);
+            }
+
+            audioStream?.Dispose();
+
             if (formatContext is not null)
             {
                 ffmpeg.avformat_close_input(&formatContext);
@@ -201,52 +265,102 @@ internal sealed unsafe class FfmpegNativeBinkFrameSource : IBinkFrameDecoder
         }
     }
 
+    private static int TryOpenAudioDecoder(
+        AVFormatContext* formatContext,
+        out AVCodecContext* codecContext,
+        out int outputSampleRate)
+    {
+        codecContext = null;
+        outputSampleRate = 0;
+
+        AVCodec* decoder = null;
+        var streamIndex = ffmpeg.av_find_best_stream(
+            formatContext, AVMediaType.AVMEDIA_TYPE_AUDIO, -1, -1, &decoder, 0);
+        if (streamIndex < 0 || decoder is null)
+        {
+            return -1;
+        }
+
+        var candidate = ffmpeg.avcodec_alloc_context3(decoder);
+        if (candidate is null)
+        {
+            return -1;
+        }
+
+        var stream = formatContext->streams[streamIndex];
+        if (ffmpeg.avcodec_parameters_to_context(candidate, stream->codecpar) < 0)
+        {
+            ffmpeg.avcodec_free_context(&candidate);
+            return -1;
+        }
+
+        candidate->thread_count = 0;
+        candidate->thread_type = ffmpeg.FF_THREAD_FRAME | ffmpeg.FF_THREAD_SLICE;
+        if (ffmpeg.avcodec_open2(candidate, decoder, null) < 0)
+        {
+            ffmpeg.avcodec_free_context(&candidate);
+            return -1;
+        }
+
+        outputSampleRate = candidate->sample_rate > 0 ? candidate->sample_rate : 48_000;
+        codecContext = candidate;
+        return streamIndex;
+    }
+
     public bool TryDecodeNextFrame(Span<byte> destination)
     {
-        var stride = checked((int)(Width * 4));
-        var required = (long)stride * Height;
-        if (destination.Length < required)
+        lock (_decodeGate)
         {
-            return false;
-        }
+            if (Volatile.Read(ref _disposed) != 0)
+            {
+                return false;
+            }
 
-        if (!TryReceiveFrame())
-        {
-            return false;
-        }
+            var stride = checked((int)(Width * 4));
+            var required = (long)stride * Height;
+            if (destination.Length < required)
+            {
+                return false;
+            }
 
-        _swsContext = ffmpeg.sws_getCachedContext(
-            _swsContext,
-            _frame->width,
-            _frame->height,
-            (AVPixelFormat)_frame->format,
-            (int)Width,
-            (int)Height,
-            AVPixelFormat.AV_PIX_FMT_BGRA,
-            ffmpeg.SWS_FAST_BILINEAR,
-            null,
-            null,
-            null);
-        if (_swsContext is null)
-        {
-            ffmpeg.av_frame_unref(_frame);
-            return false;
-        }
+            if (!TryReceiveFrame())
+            {
+                return false;
+            }
 
-        fixed (byte* destinationPointer = destination)
-        {
-            var destinationPlanes = new byte*[4] { destinationPointer, null, null, null };
-            var destinationStrides = new int[4] { stride, 0, 0, 0 };
-            var convertedRows = ffmpeg.sws_scale(
+            _swsContext = ffmpeg.sws_getCachedContext(
                 _swsContext,
-                _frame->data,
-                _frame->linesize,
-                0,
+                _frame->width,
                 _frame->height,
-                destinationPlanes,
-                destinationStrides);
-            ffmpeg.av_frame_unref(_frame);
-            return convertedRows == (int)Height;
+                (AVPixelFormat)_frame->format,
+                (int)Width,
+                (int)Height,
+                AVPixelFormat.AV_PIX_FMT_BGRA,
+                ffmpeg.SWS_FAST_BILINEAR,
+                null,
+                null,
+                null);
+            if (_swsContext is null)
+            {
+                ffmpeg.av_frame_unref(_frame);
+                return false;
+            }
+
+            fixed (byte* destinationPointer = destination)
+            {
+                var destinationPlanes = new byte*[4] { destinationPointer, null, null, null };
+                var destinationStrides = new int[4] { stride, 0, 0, 0 };
+                var convertedRows = ffmpeg.sws_scale(
+                    _swsContext,
+                    _frame->data,
+                    _frame->linesize,
+                    0,
+                    _frame->height,
+                    destinationPlanes,
+                    destinationStrides);
+                ffmpeg.av_frame_unref(_frame);
+                return convertedRows == (int)Height;
+            }
         }
     }
 
@@ -291,9 +405,17 @@ internal sealed unsafe class FfmpegNativeBinkFrameSource : IBinkFrameDecoder
             {
                 _draining = true;
                 ffmpeg.avcodec_send_packet(_codecContext, null);
+                DrainAudioDecoder();
                 return true;
             }
 
+            if (_packet->stream_index == _audioStreamIndex)
+            {
+                DecodeAudioPacket(_packet);
+                ffmpeg.av_packet_unref(_packet);
+                continue;
+            }
+
             if (_packet->stream_index != _videoStreamIndex)
             {
                 ffmpeg.av_packet_unref(_packet);
@@ -311,6 +433,248 @@ internal sealed unsafe class FfmpegNativeBinkFrameSource : IBinkFrameDecoder
         }
     }
 
+    private void DecodeAudioPacket(AVPacket* packet)
+    {
+        if (_audioCodecContext is null || _audioFrame is null || _audioFailed)
+        {
+            return;
+        }
+
+        var sendResult = ffmpeg.avcodec_send_packet(_audioCodecContext, packet);
+        if (sendResult == ffmpeg.AVERROR(ffmpeg.EAGAIN))
+        {
+            DrainAvailableAudioFrames();
+            sendResult = ffmpeg.avcodec_send_packet(_audioCodecContext, packet);
+        }
+
+        if (sendResult < 0)
+        {
+            DisableAudio("packet decode failed");
+            return;
+        }
+
+        DrainAvailableAudioFrames();
+    }
+
+    private void DrainAudioDecoder()
+    {
+        if (_audioCodecContext is null || _audioFrame is null ||
+            _audioDraining || _audioFailed)
+        {
+            return;
+        }
+
+        _audioDraining = true;
+        var sendResult = ffmpeg.avcodec_send_packet(_audioCodecContext, null);
+        if (sendResult >= 0 || sendResult == ffmpeg.AVERROR(ffmpeg.EAGAIN))
+        {
+            DrainAvailableAudioFrames();
+        }
+    }
+
+    private void DrainAvailableAudioFrames()
+    {
+        while (_audioCodecContext is not null && _audioFrame is not null)
+        {
+            var receiveResult = ffmpeg.avcodec_receive_frame(_audioCodecContext, _audioFrame);
+            if (receiveResult == ffmpeg.AVERROR(ffmpeg.EAGAIN) ||
+                receiveResult == ffmpeg.AVERROR_EOF)
+            {
+                return;
+            }
+
+            if (receiveResult < 0)
+            {
+                DisableAudio("frame decode failed");
+                return;
+            }
+
+            if (!SubmitAudioFrame())
+            {
+                ffmpeg.av_frame_unref(_audioFrame);
+                DisableAudio("host submission failed");
+                return;
+            }
+
+            ffmpeg.av_frame_unref(_audioFrame);
+        }
+    }
+
+    private bool SubmitAudioFrame()
+    {
+        if (_audioStream is null || _audioFrame is null ||
+            _audioFrame->nb_samples <= 0 || _audioFrame->extended_data is null)
+        {
+            return true;
+        }
+
+        var sampleRate = _audioFrame->sample_rate > 0
+            ? _audioFrame->sample_rate
+            : _audioCodecContext->sample_rate;
+        if (sampleRate <= 0)
+        {
+            return false;
+        }
+
+        var inputLayout = _audioFrame->ch_layout;
+        var ownsInputLayout = false;
+        if (ffmpeg.av_channel_layout_check(&inputLayout) == 0)
+        {
+            inputLayout = _audioCodecContext->ch_layout;
+        }
+        if (ffmpeg.av_channel_layout_check(&inputLayout) == 0)
+        {
+            ffmpeg.av_channel_layout_default(
+                &inputLayout,
+                Math.Max(1, _audioFrame->ch_layout.nb_channels));
+            ownsInputLayout = true;
+        }
+
+        try
+        {
+            if (!EnsureAudioResampler(
+                    &inputLayout,
+                    (AVSampleFormat)_audioFrame->format,
+                    sampleRate))
+            {
+                return false;
+            }
+
+            var maximumSamples = ffmpeg.swr_get_out_samples(
+                _swrContext, _audioFrame->nb_samples);
+            if (maximumSamples <= 0)
+            {
+                return true;
+            }
+
+            var outputBytes = checked(
+                maximumSamples * OutputAudioChannels * OutputAudioBytesPerSample);
+            var buffer = ArrayPool<byte>.Shared.Rent(outputBytes);
+            try
+            {
+                fixed (byte* output = buffer)
+                {
+                    var outputPlanes = stackalloc byte*[1];
+                    outputPlanes[0] = output;
+                    var convertedSamples = ffmpeg.swr_convert(
+                        _swrContext,
+                        outputPlanes,
+                        maximumSamples,
+                        _audioFrame->extended_data,
+                        _audioFrame->nb_samples);
+                    if (convertedSamples < 0)
+                    {
+                        return false;
+                    }
+
+                    var convertedBytes = checked(
+                        convertedSamples * OutputAudioChannels * OutputAudioBytesPerSample);
+                    return _audioStream.Submit(buffer.AsSpan(0, convertedBytes));
+                }
+            }
+            finally
+            {
+                ArrayPool<byte>.Shared.Return(buffer);
+            }
+        }
+        finally
+        {
+            if (ownsInputLayout)
+            {
+                ffmpeg.av_channel_layout_uninit(&inputLayout);
+            }
+        }
+    }
+
+    private bool EnsureAudioResampler(
+        AVChannelLayout* inputLayout,
+        AVSampleFormat inputFormat,
+        int inputSampleRate)
+    {
+        var storedInputLayout = _swrInputLayout;
+        if (_swrContext is not null &&
+            _swrInputFormat == inputFormat &&
+            _swrInputSampleRate == inputSampleRate &&
+            ffmpeg.av_channel_layout_compare(&storedInputLayout, inputLayout) == 0)
+        {
+            return true;
+        }
+
+        FreeAudioResampler();
+
+        AVChannelLayout copiedInputLayout = default;
+        if (ffmpeg.av_channel_layout_copy(&copiedInputLayout, inputLayout) < 0)
+        {
+            return false;
+        }
+
+        AVChannelLayout outputLayout = default;
+        ffmpeg.av_channel_layout_default(&outputLayout, OutputAudioChannels);
+        SwrContext* context = null;
+        var allocateResult = ffmpeg.swr_alloc_set_opts2(
+            &context,
+            &outputLayout,
+            AVSampleFormat.AV_SAMPLE_FMT_S16,
+            _audioOutputSampleRate,
+            &copiedInputLayout,
+            inputFormat,
+            inputSampleRate,
+            0,
+            null);
+        ffmpeg.av_channel_layout_uninit(&outputLayout);
+        if (allocateResult < 0 || context is null || ffmpeg.swr_init(context) < 0)
+        {
+            if (context is not null)
+            {
+                ffmpeg.swr_free(&context);
+            }
+            ffmpeg.av_channel_layout_uninit(&copiedInputLayout);
+            return false;
+        }
+
+        _swrContext = context;
+        _swrInputLayout = copiedInputLayout;
+        _swrInputLayoutValid = true;
+        _swrInputFormat = inputFormat;
+        _swrInputSampleRate = inputSampleRate;
+        return true;
+    }
+
+    private void DisableAudio(string reason)
+    {
+        if (_audioFailed)
+        {
+            return;
+        }
+
+        _audioFailed = true;
+        Console.Error.WriteLine($"[LOADER][WARN] Bink audio disabled: {reason}.");
+        FreeAudioResampler();
+        _audioStream?.Dispose();
+        _audioStream = null;
+    }
+
+    private void FreeAudioResampler()
+    {
+        if (_swrContext is not null)
+        {
+            var context = _swrContext;
+            ffmpeg.swr_free(&context);
+            _swrContext = null;
+        }
+
+        if (_swrInputLayoutValid)
+        {
+            var inputLayout = _swrInputLayout;
+            ffmpeg.av_channel_layout_uninit(&inputLayout);
+            _swrInputLayout = default;
+            _swrInputLayoutValid = false;
+        }
+
+        _swrInputFormat = AVSampleFormat.AV_SAMPLE_FMT_NONE;
+        _swrInputSampleRate = 0;
+    }
+
     public void Dispose()
     {
         if (Interlocked.Exchange(ref _disposed, 1) != 0)
@@ -318,38 +682,59 @@ internal sealed unsafe class FfmpegNativeBinkFrameSource : IBinkFrameDecoder
             return;
         }
 
-        if (_swsContext is not null)
+        lock (_decodeGate)
         {
-            ffmpeg.sws_freeContext(_swsContext);
-            _swsContext = null;
-        }
+            FreeAudioResampler();
+            _audioStream?.Dispose();
+            _audioStream = null;
 
-        if (_packet is not null)
-        {
-            var packet = _packet;
-            ffmpeg.av_packet_free(&packet);
-            _packet = null;
-        }
+            if (_swsContext is not null)
+            {
+                ffmpeg.sws_freeContext(_swsContext);
+                _swsContext = null;
+            }
 
-        if (_frame is not null)
-        {
-            var frame = _frame;
-            ffmpeg.av_frame_free(&frame);
-            _frame = null;
-        }
+            if (_packet is not null)
+            {
+                var packet = _packet;
+                ffmpeg.av_packet_free(&packet);
+                _packet = null;
+            }
 
-        if (_codecContext is not null)
-        {
-            var codecContext = _codecContext;
-            ffmpeg.avcodec_free_context(&codecContext);
-            _codecContext = null;
-        }
+            if (_frame is not null)
+            {
+                var frame = _frame;
+                ffmpeg.av_frame_free(&frame);
+                _frame = null;
+            }
 
-        if (_formatContext is not null)
-        {
-            var formatContext = _formatContext;
-            ffmpeg.avformat_close_input(&formatContext);
-            _formatContext = null;
+            if (_audioFrame is not null)
+            {
+                var frame = _audioFrame;
+                ffmpeg.av_frame_free(&frame);
+                _audioFrame = null;
+            }
+
+            if (_codecContext is not null)
+            {
+                var codecContext = _codecContext;
+                ffmpeg.avcodec_free_context(&codecContext);
+                _codecContext = null;
+            }
+
+            if (_audioCodecContext is not null)
+            {
+                var codecContext = _audioCodecContext;
+                ffmpeg.avcodec_free_context(&codecContext);
+                _audioCodecContext = null;
+            }
+
+            if (_formatContext is not null)
+            {
+                var formatContext = _formatContext;
+                ffmpeg.avformat_close_input(&formatContext);
+                _formatContext = null;
+            }
         }
     }
 }