tensorflow
librosa
numpy
gradio
soundfile
audioread