Google offers AI-powered music generation tools that let anyone create original music tracks from text descriptions or images using its Lyria AI model, available in the Gemini app, Google Vids, and through the Gemini API. Lyria 3 launched in February 2026 for 30-second clips, and Lyria 3 Pro launched in March 2026 enabling tracks up to three minutes long.

Google's music generation technology evolved from MusicLM (2023) through MusicFX and Lyria 2 to Lyria 3 (February 2026) and Lyria 3 Pro (March 25, 2026). Lyria 3 generates 30-second tracks with custom lyrics, tempo controls, and vocals from text or image prompts. Lyria 3 Pro generates tracks up to three minutes with full song structure awareness (intros, verses, choruses, bridges) and is available to paid Gemini subscribers, Google Workspace customers, Vertex AI enterprise users, and via the Gemini API. All tracks are embedded with SynthID watermarks. Google states it trained these models on partner-licensed data and permissible YouTube content, and the models do not directly mimic specific artists.