v0.9.0 is out: discli can now join Discord voice calls and transcribe them live
by•
Hey everyone, posting an update to anyone who followed the launch.
v0.9.0 just shipped and it's the release I'm most excited about. The headline: discli can now sit in a Discord voice channel, transcribe every speaker live with their display names, and stream the transcript to disk or to your AI agent in real time. Meeting notes that write themselves, from the terminal.
- GitHub release: https://github.com/DevRohit06/discli/releases/tag/v0.9.0
- Install: `pip install -U discord-cli-agent`
- Voice guide: https://discli.rohitk06.in/guides/voice
- Meeting transcription walkthrough: https://discli.rohitk06.in/use-cases/meeting-transcription
- `discli doctor` reference: https://discli.rohitk06.in/reference/cli-commands
40 views
Replies
Live transcription directly to the terminal? That is a dev’s dream. No more 'wait, what did the PM just say?' panic. How are you handling speaker diarization in noisy channels? If it can tell three people talking at once apart, you’ve basically built a superpower.
@rivra_dev Ha, thanks. Honest answer: I'm cheating on diarization. Discord's voice gateway hands you per-user PCM streams via voice_recv, each one tagged with the speaker's user_id. discli just opens a separate Deepgram (or Whisper) session per speaker, so three people talking at once isn't a cocktail party problem, it's three parallel websockets. Transcript events come out interleaved by timestamp with user_id attached.
Where it breaks: a hot mic picking up someone else in the room. That genuinely is two voices on one stream and discli treats it as one speaker. Haven't tackled that yet, the per-user channel covers most meeting cases cleanly.