privacy. precision-2 runs on pyannoteAI servers, so your audio leaves your machine.
cost. Free credits run out; check pyannoteAI pricing before processing a large archive.
Who it is forA team comparing open-source quality against a hosted service.
The problemHard audio with many speakers can need a stronger model than you can host.
The outcomeSpeaker turns from the precision-2 service, printed the same way as the local pipeline.
What you need first
- ffmpeg installed (needed by torchcodec audio decoding)
- Python with `pip install pyannote.audio` or `uv add pyannote.audio`
- A Hugging Face access token and accepted user conditions for pyannote/speaker-diarization-community-1; a GPU is optional
Steps
- Install with
pip install pyannote.audio - Create a pyannoteAI API key at dashboard.pyannote.ai, which comes with free credits
- Load Pipeline.from_pretrained("pyannote/speaker-diarization-precision-2", token=...) with that key
- Run it on audio.wav and loop over output.speaker_diarization, then compare the turns with your community-1 run
Prompt for your agent
With pyannote/pyannote-audio, run both community-1 locally and precision-2 with my PYANNOTEAI_API_KEY on <one test recording>, and show me where the speaker turns differ.
Caveat. Speaker names come back as SPEAKER_00 style labels from the premium service.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for pyannote/pyannote-audio and "Try the premium precision-2 pipeline"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer