this is awesome I was already trying to build the exact same thing but using Orange Pi 5 and Whisper speech to text model.