transformers
datasets
soundfile
accelerate
speechbrain==0.5.16
IPython