pyannote.audio

pyannote.audio

pyannote.audio is the leading open-source toolkit for speaker diarisation and voice activity detection (VAD). It provides pre-trained neural networks and a pipeline abstraction to turn raw audio into "who spoke when" annotations, powering applications from meeting transcription to media indexing.

Key Features

• State-of-the-art speaker diarisation pipeline (pyannote/speaker-diarization-3.1)
• Voice Activity Detection (pyannote/voice-activity-detection)
• Overlapped Speech Detection (OSD) and speaker segmentation
• Tight integration with HuggingFace Hub for model loading
• Integrates with OpenAI Whisper for combined transcription + attribution

Quick Start

pip install pyannote.audio

from pyannote.audio import Pipeline

pipeline = Pipeline.from_pretrained(
    "pyannote/speaker-diarization-3.1",
    use_auth_token="YOUR_HF_TOKEN",  # accept terms on HF Hub
)
diarization = pipeline("meeting.wav")
for turn, _, speaker in diarization.itertracks(yield_label=True):
    print(f"{turn.start:.1f}s - {turn.end:.1f}s: {speaker}")

npx ai-supply add pyannote-speaker-diarization

Curated mirror of the open-source pyannote.audio (MIT). Get it from the source.