How to Use Discord Bot Voice Recording With Pycord Recording Module

Discord bots can record voice from voice channels using the Pycord library’s built-in recording module. This feature lets your bot capture audio from specific users, which is useful for meeting logs, podcast archiving, or transcription. The Pycord recording module handles the WebSocket and audio data streams automatically, so you do not need to manage raw Opus encoding or RTP packets. This article explains how to set up a Pycord bot, join a voice channel, start and stop recording, and save audio files to disk.

Overview of the Pycord Recording Module

Pycord is a modern Python wrapper for the Discord API. Version 2.0 and above includes a voice recording module that lets bots receive audio from voice channels. The module uses the VoiceRecvClient class, which extends the standard voice client with listening capabilities. When the bot joins a voice channel, it can listen to all users who are speaking. The recorded audio is delivered as Opus packets. Pycord provides several built-in sinks that convert these Opus packets into common audio formats such as WAV and MP3. You can also write a custom sink if you need a different output format.

Before you start, you need a Discord bot token, Python 3.8 or later, and Pycord installed. Install Pycord with the voice extra: pip install py-cord[voice]. The bot must have the voice_states and guilds intents enabled. You can enable them in the Discord Developer Portal under the Bot section. The recording feature works in server voice channels. It does not work in group DMs or stage channels.

Steps to Record Voice With a Pycord Bot

Follow these steps to create a bot that joins a voice channel, records audio, and saves it to a file. The example uses the default sink, which outputs raw PCM data. For WAV or MP3 output, replace the sink with the appropriate class.

1. Create the bot and enable intents
   Create a new Python file. Import discord and create a Bot instance with the voice_states and guilds intents. Set the command prefix. Example: bot = discord.Bot(intents=discord.Intents.all()). Using all intents is the simplest approach for development.
2. Write a command to join a voice channel
   Create a slash command or prefix command that takes a voice channel as an argument. Inside the command, call await channel.connect(cls=discord.VoiceRecvClient). This connects the bot and returns a VoiceRecvClient instance. Store the client in a variable for later use.
3. Start recording audio
   Call voice_client.listen(discord.sinks.WaveSink()) to start recording. The WaveSink saves audio as WAV files. You can pass discord.sinks.MP3Sink() for MP3 output. To record only one user, pass the user’s ID as a second argument: voice_client.listen(sink, user_id). The bot will record only that user’s audio.
4. Stop recording and save the files
   Call voice_client.stop_listening() to stop the recording. After stopping, the sink object contains the recorded data. Access the data using sink.audio_data, which is a dictionary mapping user IDs to their audio data. Write each user’s audio to a file. Example: with open(f"{user_id}.wav", "wb") as f: f.write(sink.audio_data[user_id].file.read()).
5. Leave the voice channel
   After saving the files, call await voice_client.disconnect() to leave the channel and free resources.

Common Mistakes and Limitations to Avoid

Bot does not receive audio after joining

This usually happens when the bot lacks the necessary intents. Make sure you enabled voice_states in the Discord Developer Portal and passed the correct intents when creating the Bot instance. Also check that the bot has the connect and speak permissions in the server.

Recorded audio is silent or has low volume

The default sink records audio at the original volume from Discord. If the file is silent, the user may not have been speaking. Use a sink that writes audio immediately to disk to verify that data is being captured. Also check that the sink’s encoding matches your player. For example, WaveSink outputs 16-bit 48 kHz PCM. If your player expects a different format, you may need to convert the file using a library like pydub.

Memory usage grows too high during long recordings

The default sinks store all audio in memory until you call stop_listening. For long recordings, this can consume a lot of RAM. Use a custom sink that writes audio to disk in real time. To create a custom sink, subclass discord.sinks.Sink and implement the write method to write audio data to a file as it arrives.

Comparison of Pycord Audio Sinks

Item	WaveSink	MP3Sink	DefaultSink
Output format	WAV (16-bit PCM)	MP3	Raw PCM
File extension	.wav	.mp3	None (raw bytes)
Requires FFmpeg	No	Yes	No
Memory usage	Stores all audio in memory	Stores all audio in memory	Stores all audio in memory
Best for	Quick testing, archival	Smaller file size, sharing	Custom processing

Pycord’s recording module is a reliable way to capture voice from Discord voice channels. You now know how to set up a bot, join a voice channel, start and stop recording with the listen and stop_listening methods, and save audio files using sinks. For production use, consider writing a custom sink that writes to disk continuously to avoid memory issues. The discord.sinks module also supports stereo recording and per-user files, which you can configure through the sink parameters.