torch
numpy
soundfile
transformers
sklearn
sklearn.model_selection