pyannote.audio
Build local speaker diarization, speech-activity and speaker-change pipelines in Python.
What’s included for free?
The Python toolkit is free from source. Some pretrained pipelines require accepting model terms and a Hugging Face token; compute and hosting are separate.
Where it runs: Local install
Provider information ↗Checked 2026-10-30
Before you use it
MIT for the toolkit; pretrained models and datasets retain separate terms.
Provider terms can change. Check current pricing and usage rights before starting a project.