gradio>=5.0.0
huggingface_hub<0.26.0
datasets>=2.14.0
pandas>=2.0.0
audioop-lts