Gemini adds 30-second music generation with Lyria 3

Google started rolling out Lyria 3 in the Gemini app on 18 February, allowing users to generate 30-second tracks from a text description or uploaded image. The model can generate lyrics as well as music, with controls for style, vocals and tempo. Initial language support included English and Hindi among eight languages.

Access was for users aged 18 and over, with desktop rollout first and mobile following over several days. Paid Google AI subscribers received higher limits. Google said all tracks carried SynthID and expanded Gemini’s verification tools to audio. It also acknowledged that copyright protections were not foolproof. A watermark helps identify origin; it does not itself establish that a track is cleared for every commercial use.