Helping llms hear and speak @google deepmind.

Native Audio in Gemini 1.5. Previously Universal Speech Model (USM), Next thousand languages on Translate, Massive massively multilingual models