numpy
torch
torchaudio
tqdm
openai-whisper<=20240930,>=20230314

[fw]
faster-whisper

[hf]
transformers>=4.23.0
optimum
accelerate
