How to fetch audio segments in real time from diarization pipeline

[Yagna24]

Is there a way around to get the discrete audio segments from the diarization. I need to store the segments.

[Anticope12]

Hi @Yagna24 , Were you able to solve this ?

[Yagna24]

Yes @Anticope12 , what you can do is,

Load and apply diarization

pipeline = torch.hub.load('pyannote/pyannote-audio', 'dia_ami')
diarized = pipeline(audio_input)

Get contents out of the diarized audio that contains the timestamps and labels

segments = diarized.for_json()["content"]