torch
torchaudio
funasr>=1.2.0
modelscope
huggingface_hub
moviepy
gradio
numpy<2.0
librosa
soundfile