4 ms·Mini-Omni: Language Models Can Hear, Talk While Thinking in Streaming3 points by thenameless7741 2y ago