Kyutai's Speech-To-Text and Text-To-Speech models based on the Delayed Streams Modeling framework. - View it on GitHub