kyutai-labs/moshi

Moshi is a speech-text foundation model and full-duplex spoken dialogue framework. It uses Mimi, a state-of-the-art streaming neural audio codec.

拖动卡牌调整角度，点击翻面查看卡背。