voice-cat/clients/apple/dotnet/VoiceCat.iOS/IosAudioEngine.cs

using System.Runtime.InteropServices;
using AVFoundation;
using VoiceCat.Audio;
using VoiceCat.Core;

namespace VoiceCat.iOS;

internal sealed class IosAudioEngine
{
    internal static IosAudioEngine Shared { get; } = new();
    private readonly PcmRing playbackRing = new(131_072);
    private readonly short[] renderScratch = new short[16_384];
    private AVAudioEngine? engine;
    private AVAudioSourceNode? source;
    private AVAudioFormat? outputFormat;
    private VoiceCatClient? client;
    private uint microphoneStream;
    private int microphoneChannels = 1;
    private readonly PcmRing microphoneRing = new(131_072);
    private readonly short[] microphoneFrame = new short[960 * 2];
    private readonly CancellationTokenSource microphoneStop = new();
    private readonly Task microphoneWorker;
    private AVAudioFormat? microphoneFormat;
    private AVAudioConverter? microphoneConverter;
    private AVAudioPcmBuffer? convertedMicrophone;
    private AVAudioPcmBuffer? pendingInput;
    private AVAudioConverterInputHandler? inputProvider;
    private bool inputProvided;
    private bool tapInstalled;
    internal bool IsConnected { get; private set; }

    private IosAudioEngine() { microphoneWorker = PumpMicrophoneAsync(microphoneStop.Token); }

    internal void StartListening(VoiceCatClient owner)
    {
        Stop(); client = owner; IsConnected = true; owner.Audio.MixedPcm += ReceiveMixedPcm; Rebuild();
    }

    internal void StartMicrophone(uint streamId, int channels)
    {
        microphoneStream = streamId; microphoneChannels = Math.Clamp(channels, 1, 2); Rebuild();
    }

    internal void StopMicrophone() { microphoneStream = 0; Rebuild(); }
    internal void Reconfigure() { if (IsConnected) Rebuild(); }

    private void Rebuild()
    {
        DestroyGraph(); IosAudioRouter.Shared.Apply();
        var next = new AVAudioEngine();
        outputFormat = new(AVAudioCommonFormat.PCMFloat32, 48_000, 2, false);
        source = new(outputFormat, Render);
        next.AttachNode(source);
        NSError? connectionError = null;
        if (OperatingSystem.IsIOSVersionAtLeast(27)) next.Connect(source, next.MainMixerNode, outputFormat, out connectionError);
        else next.Connect(source, next.MainMixerNode, outputFormat);
        if (connectionError is not null) throw new InvalidOperationException(connectionError.LocalizedDescription);
        if (microphoneStream != 0)
        {
            AVAudioInputNode input = next.InputNode;
            input.SetVoiceProcessingEnabled(IosAudioRouter.Shared.UsesVoiceProcessing, out _);
            if (IosAudioRouter.Shared.UsesVoiceProcessing) input.VoiceProcessingAgcEnabled = IosAudioRouter.Shared.AutomaticGainControl;
            AVAudioFormat inputFormat = input.GetBusOutputFormat(0);
            microphoneFormat = new(AVAudioCommonFormat.PCMInt16, 48_000, (uint)microphoneChannels, true);
            microphoneConverter = new(inputFormat, microphoneFormat);
            uint capacity = checked((uint)Math.Ceiling(4_096 * 48_000 / inputFormat.SampleRate) + 64);
            convertedMicrophone = new(microphoneFormat, capacity);
            inputProvider = ProvideInput;
            NSError? tapError = null;
            if (OperatingSystem.IsIOSVersionAtLeast(27)) input.InstallTapOnBus(0, 960, inputFormat, out tapError, Capture);
            else input.InstallTapOnBus(0, 960, inputFormat, Capture);
            if (tapError is not null) throw new InvalidOperationException(tapError.LocalizedDescription);
            tapInstalled = true;
        }
        next.Prepare();
        if (!next.StartAndReturnError(out NSError? error)) { next.Dispose(); throw new InvalidOperationException(error.LocalizedDescription); }
        engine = next;
    }

    private unsafe void Capture(AVAudioPcmBuffer buffer, AVAudioTime time)
    {
        VoiceCatClient? owner = client; uint stream = microphoneStream;
        if (owner is null || stream == 0 || buffer.FrameLength == 0) return;
        AVAudioConverter? converter = microphoneConverter;
        AVAudioPcmBuffer? converted = convertedMicrophone;
        AVAudioConverterInputHandler? provider = inputProvider;
        if (converter is null || converted is null || provider is null) return;
        pendingInput = buffer; inputProvided = false; converted.FrameLength = 0;
        converter.ConvertToBuffer(converted, out _, provider);
        if (converted.FrameLength == 0) return;
        nint samples = Marshal.ReadIntPtr(converted.Int16ChannelData);
        if (samples != 0) microphoneRing.TryWrite(new ReadOnlySpan<short>((void*)samples, checked((int)converted.FrameLength * microphoneChannels)));
        pendingInput = null;
    }

    private AVAudioBuffer ProvideInput(uint _, out AVAudioConverterInputStatus status)
    {
        if (!inputProvided && pendingInput is { } input) { inputProvided = true; status = AVAudioConverterInputStatus.HaveData; return input; }
        status = AVAudioConverterInputStatus.NoDataNow; return null!;
    }

    private async Task PumpMicrophoneAsync(CancellationToken token)
    {
        using var timer = new PeriodicTimer(TimeSpan.FromMilliseconds(10));
        while (await timer.WaitForNextTickAsync(token).ConfigureAwait(false))
        {
            int channels = microphoneChannels, required = 960 * channels;
            while (microphoneRing.Count >= required)
            {
                int read = microphoneRing.Read(microphoneFrame.AsSpan(0, required));
                VoiceCatClient? owner = client; uint stream = microphoneStream;
                if (read == required && owner is not null && stream != 0) owner.Audio.FeedPcm(stream, microphoneFrame.AsSpan(0, required), channels);
            }
        }
    }

    private void ReceiveMixedPcm(ReadOnlySpan<short> pcm) => playbackRing.TryWrite(pcm);

    private unsafe int Render(IntPtr isSilence, IntPtr timestamp, uint frameCount, IntPtr outputData)
    {
        int frames = checked((int)frameCount), requested = checked(frames * 2);
        if (requested > renderScratch.Length) return -1;
        Span<short> input = renderScratch.AsSpan(0, requested);
        int read = playbackRing.Read(input); input[read..].Clear();
        int count = Marshal.ReadInt32(outputData), first = IntPtr.Size == 8 ? 8 : 4, stride = IntPtr.Size == 8 ? 16 : 12;
        if (count != 2) return -1;
        for (int channel = 0; channel < 2; channel++)
        {
            nint data = Marshal.ReadIntPtr(outputData, first + channel * stride + 8);
            var output = new Span<float>((void*)data, frames);
            for (int frame = 0; frame < frames; frame++) output[frame] = input[frame * 2 + channel] / 32768f;
        }
        if (isSilence != IntPtr.Zero) Marshal.WriteByte(isSilence, read == 0 ? (byte)1 : (byte)0);
        return 0;
    }

    internal void Stop()
    {
        IsConnected = false; microphoneStream = 0;
        if (client is { } owner) owner.Audio.MixedPcm -= ReceiveMixedPcm;
        DestroyGraph(); client = null; IosAudioRouter.Shared.Deactivate();
    }

    private void DestroyGraph()
    {
        if (engine is { } old)
        {
            if (tapInstalled) old.InputNode.RemoveTapOnBus(0);
            old.Stop(); if (source is not null) old.DetachNode(source); old.Dispose();
        }
        tapInstalled = false; pendingInput = null; inputProvider = null;
        convertedMicrophone?.Dispose(); convertedMicrophone = null;
        microphoneConverter?.Dispose(); microphoneConverter = null;
        microphoneFormat?.Dispose(); microphoneFormat = null;
        source?.Dispose(); source = null; outputFormat?.Dispose(); outputFormat = null; engine = null;
    }
}