transformers>=4.25.0
torch
speechbox>=0.1.2
librosa
accelerate
pyannote.audio