fastrtc[vad]
groq
edge-tts
librosa
numpy
soundfile