Open prkumar112451 opened 6 months ago
I am having the same problem since the most recent update. Transcription is still quick and Diarization used to take the same amount of time (roughly).
@prkumar112451 You consider slow 45 seconds for 20 minutes of audio? I'm running insanely-fast-whisper on a macOS with M1 and it takes 30 seconds to diarize 2.40 minutes of audio. For my use case it is not slow at all. Your scenario doesn't (subjectively) seem slow to me either.
Diarizing a 20 minutes of audio is taking roughly 45 seconds time. Could you please let me know if the optimizations of insanely-fast-whisper focused just on whisper or we are planning to add some optimization for diarization (pyannote) as well.