meeting-assistant/MeetingAssistant/Recording/RollingAudioBuffer.cs

using NAudio.Wave;

namespace MeetingAssistant.Recording;

internal sealed class RollingAudioBuffer
{
    private readonly TimeSpan retention;
    private readonly object gate = new();
    private readonly Queue<TimedAudioChunk> chunks = new();
    private TimeSpan position;

    public RollingAudioBuffer(TimeSpan retention)
    {
        this.retention = retention > TimeSpan.Zero ? retention : TimeSpan.FromMinutes(10);
    }

    public void Append(AudioChunk chunk)
    {
        if (chunk.Pcm.Length == 0 || chunk.SampleRate <= 0 || chunk.Channels <= 0)
        {
            return;
        }

        lock (gate)
        {
            var start = position;
            var end = start + chunk.Duration;
            chunks.Enqueue(new TimedAudioChunk(start, end, chunk));
            position = end;
            Prune(end - retention);
        }
    }

    public void Reset()
    {
        lock (gate)
        {
            chunks.Clear();
            position = TimeSpan.Zero;
        }
    }

    public byte[] TryExtractWav(TimeSpan start, TimeSpan end)
    {
        if (end <= start)
        {
            return [];
        }

        List<TimedAudioChunk> matchingChunks;
        lock (gate)
        {
            matchingChunks = chunks
                .Where(chunk => chunk.End > start && chunk.Start < end)
                .ToList();
        }

        if (matchingChunks.Count == 0)
        {
            return [];
        }

        var format = new WaveFormat(
            matchingChunks[0].Chunk.SampleRate,
            16,
            matchingChunks[0].Chunk.Channels);
        using var output = new MemoryStream();
        using (var writer = new WaveFileWriter(output, format))
        {
            foreach (var timedChunk in matchingChunks)
            {
                if (timedChunk.Chunk.SampleRate != format.SampleRate || timedChunk.Chunk.Channels != format.Channels)
                {
                    continue;
                }

                var blockAlign = format.BlockAlign;
                var copyStart = Max(start, timedChunk.Start);
                var copyEnd = Min(end, timedChunk.End);
                var firstFrame = (int)Math.Floor((copyStart - timedChunk.Start).TotalSeconds * format.SampleRate);
                var afterLastFrame = (int)Math.Ceiling((copyEnd - timedChunk.Start).TotalSeconds * format.SampleRate);
                var byteOffset = Clamp(firstFrame * blockAlign, 0, timedChunk.Chunk.Pcm.Length);
                var byteCount = Clamp(afterLastFrame * blockAlign, byteOffset, timedChunk.Chunk.Pcm.Length) - byteOffset;
                if (byteCount > 0)
                {
                    writer.Write(timedChunk.Chunk.Pcm, byteOffset, byteCount);
                }
            }
        }

        return output.ToArray();
    }

    private void Prune(TimeSpan cutoff)
    {
        while (chunks.Count > 0 && chunks.Peek().End < cutoff)
        {
            chunks.Dequeue();
        }
    }

    private static TimeSpan Max(TimeSpan left, TimeSpan right)
    {
        return left >= right ? left : right;
    }

    private static TimeSpan Min(TimeSpan left, TimeSpan right)
    {
        return left <= right ? left : right;
    }

    private static int Clamp(int value, int min, int max)
    {
        return Math.Min(Math.Max(value, min), max);
    }

    private sealed record TimedAudioChunk(TimeSpan Start, TimeSpan End, AudioChunk Chunk);
}