first base model for full-duplex conversational audio - View it on GitHub