torch>=2.1.0,<2.9
torchaudio>=2.1.0,<2.9
transformers>=4.34.0,<5
gradio>=3.41.0,<6
sounddevice>=0.4.6
soundfile>=0.12.1
scipy>=1.11.0
numpy>=1.25.0
datasets>=2.13.0
huggingface_hub>=1.24.0
pydantic==2.10.6
click<8.1
protobuf<4