Table of Contents

Receiving with Voice

1. Enable Incoming Audio

Incoming receive is opt-in.

using DisCatSharp.Voice;

client.UseVoice(new VoiceConfiguration
{
    EnableIncoming = true,
    EnableDebugLogging = false
});

2. Connect and Subscribe

using DisCatSharp.Voice;

VoiceConnection connection = await channel.ConnectAsync();
connection.VoiceReceived += OnVoiceReceived;
connection.VoicePacketDropped += OnVoicePacketDropped;

3. Handle Decoded Frames

VoiceReceived gives you decoded PCM plus metadata.

using DisCatSharp.Voice;
using DisCatSharp.Voice.EventArgs;

private static Task OnVoiceReceived(VoiceConnection _, VoiceReceiveEventArgs e)
{
    Console.WriteLine(
        $"RX user={e.User?.Id} ssrc={e.Ssrc} seq={e.Sequence} " +
        $"pcm={e.PcmData.Length} opus={e.OpusData.Length} " +
        $"missing={e.MissingFrames} conceal={e.IsConcealmentFrame}");

    // e.PcmData is raw PCM bytes for this frame.
    return Task.CompletedTask;
}

Useful fields:

  • PcmData: decoded PCM for the frame
  • OpusData: original Opus payload when available
  • AudioFormat: format of decoded PCM
  • AudioDuration: frame duration in ms
  • MissingFrames: count of gap frames before this frame
  • IsConcealmentFrame: true when frame is packet-loss concealment

4. Monitor Packet Drops

using DisCatSharp.Voice;
using DisCatSharp.Voice.EventArgs;

private static Task OnVoicePacketDropped(VoiceConnection _, VoicePacketDroppedEventArgs e)
{
    Console.WriteLine($"DROP reason={e.Reason} user={e.User?.Id} ssrc={e.Ssrc} seq={e.Sequence} detail={e.Detail}");
    return Task.CompletedTask;
}

Drop reasons include malformed RTP or extensions, out-of-order packets, DAVE media not ready, missing sender mappings or decryptors, queue overflow, and Opus decode failures.

5. Record to MP3 (ffmpeg)

If you want a listen-test artifact, you can stream decoded PCM directly into ffmpeg:

using System.Diagnostics;
using DisCatSharp.Voice;
using DisCatSharp.Voice.EventArgs;

private static Process? _ffmpeg;
private static Stream? _ffmpegIn;

public static void StartMp3Recording(string outputPath)
{
    _ffmpeg = Process.Start(new ProcessStartInfo
    {
        FileName = "ffmpeg",
        Arguments = $"-y -hide_banner -loglevel warning -f s16le -ar 48000 -ac 2 -i pipe:0 -codec:a libmp3lame -q:a 2 \"{outputPath}\"",
        RedirectStandardInput = true,
        UseShellExecute = false,
        CreateNoWindow = true
    }) ?? throw new InvalidOperationException("Failed to start ffmpeg.");

    _ffmpegIn = _ffmpeg.StandardInput.BaseStream;
}

private static async Task OnVoiceReceivedForRecording(VoiceConnection _, VoiceReceiveEventArgs e)
{
    if (_ffmpegIn is null || e.PcmData.Length == 0)
        return;

    // Preserve timeline continuity by filling packet-loss gaps with silence.
    if (e.MissingFrames > 0)
    {
        var bytesPerMs = e.AudioFormat.SampleRate * e.AudioFormat.ChannelCount * sizeof(short) / 1000;
        var silenceBytes = bytesPerMs * e.AudioDuration * e.MissingFrames;
        if (silenceBytes > 0)
            await _ffmpegIn.WriteAsync(new byte[silenceBytes]);
    }

    await _ffmpegIn.WriteAsync(e.PcmData);
}

public static async Task StopMp3RecordingAsync()
{
    if (_ffmpeg is null || _ffmpegIn is null)
        return;

    await _ffmpegIn.FlushAsync();
    _ffmpegIn.Dispose();
    _ffmpegIn = null;

    _ffmpeg.WaitForExit();
    _ffmpeg.Dispose();
    _ffmpeg = null;
}

Wire it like this:

connection.VoiceReceived += OnVoiceReceivedForRecording;
StartMp3Recording("logs/recordings/session.mp3");

The same pattern works for WAV by changing ffmpeg output arguments.

6. DAVE Decryption Notes

When DAVE is active for the channel, transport decryption and then per-user DAVE decryption happen before your handler runs.

  • You always receive plain PCM in VoiceReceiveEventArgs.PcmData.
  • IsE2eeUsableForReceive reports whether the executing receive mode can currently process media.
  • ReadyForTransition and Downgrading do not make an established receiver unusable. Remote-user decryptors are transitioned in place and retain previous encrypted-epoch ratchets for ten seconds.
  • A downgrade enables plaintext passthrough before OP23 while retaining old encrypted ratchets for ten seconds.
  • An upgrade allows plaintext for a ten-second overlap while the new encrypted mode begins.
  • A newly welcomed member can receive with the new-epoch ratchets prepared by OP30, but its local sender remains inactive until matching OP22.
  • DaveStateChanged and DaveOpcodeObserved help diagnose handshake timing.

Do not infer receive usability from DaveState == Active. For example, an established member can receive normally while the control plane is ReadyForTransition, and protocol-0 passthrough is usable while the control plane is Inactive.

7. Unsubscribe / Disconnect

connection.VoiceReceived -= OnVoiceReceived;
connection.VoicePacketDropped -= OnVoicePacketDropped;
connection.Disconnect();

See Also

Edit this page