librosa: Originally not used under style_bert_vits2/. pyannote.audio: Required only when using the "infer style vector from audio" function during inference, but it is currently almost unused. scipy: Because it was used in only one utility function that was not needed outside of training.
21 KiB
21 KiB