soundfile
librosa
tensorflow
faiss-cpu
numpy
groq
SpeechRecognition==3.8.1
torch
transformers
streamlit
tf-keras