A Discord bot that records voice channel audio and transcribes it using Whisper. Currently only supports Apple Silion.
- Record audio from Discord voice channels
- Separate audio files per speaker
- Transcribe speech using Whisper (base.en model)
- Generate summaries of transcriptions
-
Create a
.envfile with your Discord bot token:DISCORD_TOKEN=YOUR_TOKEN_HERE -
Install dependencies:
pip install -r requirements.txtbrew install libogg opus opusfile libopusenc libvorbis flac ffmpeg -
Run the bot:
python main.py
- Join a voice channel
- Type
/recordin a text channel - Speak in the voice channel
- Type
/stopto stop recording - The bot will post the transcription and summary
- Apple Silicon (M1/M2/M3) Mac
- Python 3.11
- Discord bot token
- Whisper model (auto-downloaded on first run)
main.py- Bot entry point and Discord commandstranscription.py- Audio transcription using Whispersummarization.py- Text summarization using Qwen3 modelrecordings/- Saved audio files (created automatically)