全双工语音助手：边听边说、打断、任务委派与工具执行。Full-duplex voice agent with barge-in, streaming audio, and async thinking. DuplexOmni (arXiv:2606.09186).
real-time websocket pytorch tts arxiv reference-implementation asr voice-assistant streaming-audio full-duplex multimodal turn-taking voice-agent llm-tools speech-language-model speech-interaction barge-in spoken-dialogue thinker-talker duplexomni
-
Updated
Oct 2, 2026 - Python