WTF, is NVIDIA's open-source PersonaPlex voice model for real? The model listens to you while you're speaking and starts organizing and generating responses at the same time. It supports interruptions, cutting in, and 'uh-huh' style feedback, making it closer to real human conversation rhythms. You can also control the 'character personality' through Persona Prompt and the 'voice tone and speaking style' through Voice Prompt, all done end-to-end. Had GPT look into it, and it's super suitable for: AI Customer Service / AI Sales Voice Agents / Companions Real-time conversation product prototypes https://(github.com)/NVIDIA/personaplex