Lyria 3 brings AI-generated music to Gemini

Users can control style, tempo, and vocals while quickly producing shareable tracks directly from the Gemini app.

Gemini’s new Lyria 3 model lets users generate short music tracks from prompts or images, adding another creative layer to the app’s growing multimedia toolkit.

The Gemini app has introduced Lyria 3, the latest music-generation model from Google DeepMind, enabling users to create 30-second tracks from text prompts, images, or videos. The feature is rolling out in beta, marking a further expansion of creative tools within the platform.

Users can customise genre, tempo, and vocals, while the system generates lyrics automatically when needed. Tracks include AI-generated cover art and can be shared directly, aiming to provide a simple way to produce short, personalised soundtracks rather than full compositions.

Audio created in the app is embedded with SynthID watermarking to identify AI-generated content, alongside new verification tools that allow users to check whether files were produced using Google AI.

The model is designed to produce original material rather than replicate specific artists, supported by filters and reporting mechanisms.

Availability initially covers multiple major languages for users aged 18 and over, with higher usage limits offered to premium subscribers. Lyria 3 is also being integrated into YouTube creator tools to enhance Shorts soundtracks as the rollout expands.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!