Port managed client audio and Windows application

2026-09-16 16:48:06 +02:00
parent 5a226ba543
commit 82ad4c2811
56 changed files with 2304 additions and 250 deletions
@@ -1,148 +1,85 @@
+using VoiceCat.Audio;
 using VoiceCat.Interop;

-// Owns N ProcessLoopbackCapture instances, mixes their PCM every 20 ms, and feeds
-// the result to the core via vc_stream_feed_pcm. Used for per-app audio sharing.
 namespace VoiceCat.App.Audio;

+// Capture threads own their ring producers; the mix thread alone consumes them.
 public sealed class ProcessAudioMixer : IDisposable
 {
-    private const int SampleRate   = 48000;
-    private const int FrameSamples = 960;
-    private const int Channels     = 2; // stereo; captures fall back to mono if needed
-
-    private readonly List<ProcessLoopbackCapture> _captures = [];
-
-    // Per-capture latest frame, protected by _frameLock.
-    private readonly object _frameLock = new();
-    private List<short[]>   _latestFrames = [];
-    private int             _activeChannels = Channels;
-
-    private Thread?          _mixThread;
-    private volatile bool    _running;
-    private VoiceCatClient?  _client;
-    private uint             _streamId;
-
+    private sealed class Input
+    {
+        internal readonly PcmRing Ring = new(16384);
+        private readonly short[] stereo = new short[1920];
+        internal void Feed(short[] pcm, int channels)
+        {
+            if (channels == 2) { Ring.TryWrite(pcm); return; }
+            if (channels != 1 || pcm.Length > 960) return;
+            for (int i = 0; i < pcm.Length; i++) stereo[2 * i] = stereo[2 * i + 1] = pcm[i];
+            Ring.TryWrite(stereo.AsSpan(0, pcm.Length * 2));
+        }
+    }
+    private readonly List<ProcessLoopbackCapture> captures = [];
+    private Input[] inputs = [];
+    private Thread? thread;
+    private volatile bool running;
+    private VoiceCatClient? client;
+    private uint streamId;
    public void Start(AppAudioScope scope, VoiceCatClient client, uint streamId)
    {
-        if (_running) return;
-        _client   = client;
-        _streamId = streamId;
-
+        if (running) return;
+        this.client = client; this.streamId = streamId;
        var specs = ResolveCaptures(scope);
-        if (specs.Count == 0)
+        inputs = specs.Select(_ => new Input()).ToArray();
+        try
        {
-            // nothing to capture — scope resolved to empty set
-            return;
+            for (int i = 0; i < specs.Count; i++)
+            {
+                Input input = inputs[i]; var (pid, mode) = specs[i];
+                var capture = new ProcessLoopbackCapture(pid, mode);
+                capture.PcmFrameReady += (pcm, _, channels) => input.Feed(pcm, channels);
+                captures.Add(capture);
+                if (!capture.Start()) throw new InvalidOperationException("Process audio capture could not start.");
+            }
+            if (inputs.Length == 0) return;
+            running = true;
+            thread = new Thread(MixLoop) { IsBackground = true, Name = "ProcessAudioMixer" }; thread.Start();
        }
-
-        lock (_frameLock)
-        {
-            _latestFrames   = new List<short[]>(new short[specs.Count][]);
-            _activeChannels = Channels;
-        }
-
-        for (int i = 0; i < specs.Count; i++)
-        {
-            int captureIndex = i;
-            var (pid, mode) = specs[i];
-            var cap = new ProcessLoopbackCapture(pid, mode);
-            cap.PcmFrameReady += (pcm, spc, ch) => OnCaptureFrame(captureIndex, pcm, ch);
-            _captures.Add(cap);
-        }
-
-        foreach (var c in _captures) c.Start();
-
-        _running   = true;
-        _mixThread = new Thread(MixLoop) { IsBackground = true, Name = "ProcessAudioMixer" };
-        _mixThread.Start();
+        catch { Stop(); throw; }
    }
-
    public void Stop()
    {
-        _running = false;
-        _mixThread?.Join(500);
-        foreach (var c in _captures) { c.Stop(); c.Dispose(); }
-        _captures.Clear();
+        running = false;
+        if (thread is not null && !thread.Join(5000)) throw new TimeoutException("Process audio mixer did not stop.");
+        foreach (var capture in captures) capture.Dispose();
+        captures.Clear(); inputs = []; thread = null;
    }
-
    public void Dispose() => Stop();
-
-    // ── Capture callback ──────────────────────────────────────────────────────
-
-    private void OnCaptureFrame(int index, short[] pcm, int channels)
-    {
-        lock (_frameLock)
-        {
-            // Upmix mono → stereo interleave if the capture fell back to mono.
-            if (channels == 1 && _activeChannels == 2)
-                pcm = MonoToStereo(pcm);
-            if (index < _latestFrames.Count)
-                _latestFrames[index] = pcm;
-        }
-    }
-
-    // ── Mix loop (20 ms timer) ────────────────────────────────────────────────
-
    private void MixLoop()
    {
-        // Use a target period close to 20 ms; small under-shoot avoids accumulating drift.
-        const int periodMs = 19;
-        while (_running)
+        var frame = new short[1920]; var mix = new short[1920]; var sums = new int[1920];
+        long deadline = System.Diagnostics.Stopwatch.GetTimestamp();
+        while (running)
        {
-            Thread.Sleep(periodMs);
-            if (!_running) break;
-
-            short[] mix;
-            lock (_frameLock)
+            sums.AsSpan().Clear();
+            foreach (Input input in inputs)
            {
-                int len = FrameSamples * _activeChannels;
-                mix = new short[len];
-
-                foreach (var frame in _latestFrames)
-                {
-                    if (frame == null) continue;
-                    int frameLen = Math.Min(frame.Length, len);
-                    for (int i = 0; i < frameLen; i++)
-                    {
-                        int sum = mix[i] + frame[i];
-                        mix[i] = (short)Math.Clamp(sum, short.MinValue, short.MaxValue);
-                    }
-                }
+                while (input.Ring.Count > 11520) input.Ring.Read(frame);
+                frame.AsSpan().Clear(); input.Ring.Read(frame);
+                for (int i = 0; i < frame.Length; i++) sums[i] += frame[i];
            }
-
-            _client?.StreamFeedPcm(_streamId, mix, FrameSamples, (uint)_activeChannels);
+            for (int i = 0; i < mix.Length; i++) mix[i] = (short)Math.Clamp(sums[i], short.MinValue, short.MaxValue);
+            client?.StreamFeedPcm(streamId, mix, 960, 2);
+            deadline += System.Diagnostics.Stopwatch.Frequency / 50;
+            double wait = (deadline - System.Diagnostics.Stopwatch.GetTimestamp()) * 1000.0 / System.Diagnostics.Stopwatch.Frequency;
+            if (wait > 0) Thread.Sleep((int)Math.Ceiling(wait));
+            else if (wait < -100) deadline = System.Diagnostics.Stopwatch.GetTimestamp();
        }
    }
-
-    // ── Helpers ───────────────────────────────────────────────────────────────
-
-    // Resolve the scope to the WASAPI captures to open:
-    //  OnlyApps      → one INCLUDE capture per selected process tree.
-    //  AllExceptApps → one EXCLUDE capture of the single selected process tree, which
-    //                  natively captures the whole system render mix minus that tree
-    //                  (dynamic — apps launched later are included automatically).
-    private static List<(int pid, ProcessLoopbackCapture.Mode mode)> ResolveCaptures(AppAudioScope scope)
+    private static List<(int pid, ProcessLoopbackCapture.Mode mode)> ResolveCaptures(AppAudioScope scope) => scope switch
    {
-        return scope switch
-        {
-            OnlyApps      o => o.Pids.Select(p => (p, ProcessLoopbackCapture.Mode.Include)).ToList(),
-            AllExceptApps a when a.Pids.Count > 0 =>
-                [(a.Pids[0], ProcessLoopbackCapture.Mode.Exclude)],
-            // Entire desktop minus VoiceCat itself: EXCLUDE our own process tree.
-            EntireDesktop { ExcludeSelf: true } =>
-                [(Environment.ProcessId, ProcessLoopbackCapture.Mode.Exclude)],
-            _ => [],
-        };
-    }
-
-    private static short[] MonoToStereo(short[] mono)
-    {
-        var stereo = new short[mono.Length * 2];
-        for (int i = 0; i < mono.Length; i++)
-        {
-            stereo[i * 2]     = mono[i];
-            stereo[i * 2 + 1] = mono[i];
-        }
-        return stereo;
-    }
+        OnlyApps o => o.Pids.Select(p => (p, ProcessLoopbackCapture.Mode.Include)).ToList(),
+        AllExceptApps a when a.Pids.Count > 0 => [(a.Pids[0], ProcessLoopbackCapture.Mode.Exclude)],
+        EntireDesktop { ExcludeSelf: true } => [(Environment.ProcessId, ProcessLoopbackCapture.Mode.Exclude)],
+        _ => [],
+    };
 }