discord-native-voice

A native voice extension for discord.py-self.

This package plugs into discord.py-self's normal voice connection flow and adds voice receive, video, and Go Live stream support. It is meant to feel like a drop-in replacement for the existing VoiceClient.

Key Features

• Native voice send and receive
• Video send and receive for all Discord video codecs: AV1, H265, H264, VP8, and VP9
• Go Live stream creation, watching, sending, and receiving
• Media sources for FFmpeg files, desktop capture, raw audio, and raw video
• Media sinks for callbacks, queues, WAV output, FFmpeg output, and muxed audio/video recordings
• Codec advertisement based on local FFmpeg support
• RTX/NACK support for video packet recovery
• Video simulcast support
• End-to-end encryption support with DAVE
• Compatible with nearly all existing voice sinks

Installing

Python 3.10 or higher is required.

Install discord.py-self first, then install this extension:

# Linux/macOS
python3 -m pip install -U discord.py-self <todo>

# Windows
py -3 -m pip install -U discord.py-self <todo>

Note that PyNaCl is not required when using this extension.

A Rust toolchain is required when building from source. FFmpeg is also recommended, since the built-in file, desktop, and recording helpers use it.

For development tools:

python3 -m pip install -U .[dev]

Quick Example

Connect with the native voice client:

from discord.ext.native_voice import VoiceClient

voice = await channel.connect(cls=VoiceClient, self_video=True)

Play a media file with audio and video:

from discord.ext.native_voice import FFmpegMediaSource, VoiceClient

voice = await channel.connect(cls=VoiceClient, self_video=True)

source = FFmpegMediaSource.from_file(
    'clip.mp4',
    'H264',
    width=1280,
    height=720,
    fps=30,
    bitrate=4_000_000,
)

voice.play(source)

Receive media with a callback:

from discord.ext.native_voice import MediaPacket, VoiceClient

voice = await channel.connect(cls=VoiceClient)

def on_packet(packet: MediaPacket) -> None:
    print(packet.media_type, packet.codec, packet.user_id)

voice.listen(on_packet)

Go Live Streams

Streams use a separate StreamClient, created from an existing native voice connection.

from discord.ext.native_voice import FFmpegMediaSource, StreamClient, VoiceClient

voice = await channel.connect(cls=VoiceClient)
stream = await voice.create_stream(cls=StreamClient)

# Simplified example, reality is slightly more complicated
source = FFmpegMediaSource.from_desktop(
    'H264',
    width=1920,
    height=1080,
    fps=30,
    bitrate=6_000_000,
)

stream.play(source)

To watch another user's stream:

stream = voice.streams[0]
stream_client = await stream.watch(cls=StreamClient)

Configuration

Options that affect Discord's voice negotiation should be set before connecting. Use with_config() to create a configured client class:

from discord.ext.native_voice import VoiceClient

ConfiguredVoiceClient = VoiceClient.with_config(
    rtx=False,
    udp_qos=True,
    enable_debug_stats=True,
)

voice = await channel.connect(cls=ConfiguredVoiceClient, self_video=True)

Media Helpers

The extension includes source and sink helpers for common media tasks:

• FFmpegMediaSource and FFmpegVideoSource for files, pipes, and desktop capture.
• PCMMediaSource, AudioFrameSource, and VideoFrameSource for already produced audio or video.
• QueueSink and AsyncQueueSink for application-managed receive loops.
• WaveSink, MixedWaveSink, FFmpegSink, and FFmpegMuxSink for recording.
• MultiSink, PerUserSink, and filter sinks for routing packets.

For advanced integrations, received media is exposed as MediaPacket objects with RTP metadata and parsed RTP extensions.

Docs

Docs are available at https://discord-native-voice.readthedocs.io/en/latest/index.html. Credits to discord.py for the theme.

Issues

For now, issues should be opened under this repository. Eventually, they will be migrated to the discord.py-self repository.

Links

• discord.py-self