hf/transformers/examples/pytorch/speech-pretraining/requirements.txt

datasets >= 1.12.0
torch >= 1.5
torchaudio
accelerate >= 0.12.0
librosa