av==11.0.0 einops flashy hydra-core==1.1 hydra_colorlog julius num2words numpy==1.24.4 sentencepiece spacy==3.6.1 torch torchaudio tqdm transformers==4.31.0 # need Encodec there. xformers==0.0.22 demucs librosa soundfile torchmetrics encodec protobuf torchvision torchtext pesq pystoi