diarization.py

from pyannote.audio import Pipeline
from ..config import settings

_diar_pipeline = Pipeline.from_pretrained(settings.DIAR_MODEL)

def diarize_segments(file_path: str) -> list:
    diarization = _diar_pipeline(file_path)
    segments = []
    for turn, _, speaker in diarization.itertracks(yield_label=True):
        segments.append((turn.start, turn.end, speaker))
    return segments