Music Generation Comes to Gemini
Google has rolled out Lyria 3, its latest generative music model, in beta within the Gemini app. Users can now create custom 30-second music tracks by describing an idea in text or uploading a photo. The AI generates both lyrics and instrumental audio tailored to the user's prompt, complete with custom cover art.
Key Improvements in Lyria 3
The new model advances Google's music generation capabilities in three significant ways:
- Auto-generated lyrics: Users no longer need to provide their own lyrics—Lyria 3 generates them based on the prompt
- Enhanced creative control: Users can specify style, vocals, tempo, and other musical elements
- Greater complexity: The model creates more realistic and musically sophisticated tracks
How to Use the Feature
Users can approach music creation through two methods:
- Text-to-track: Describe a genre, mood, memory, or concept (e.g., "nostalgic afrobeat track about my mother's home cooking")
- Image/video-to-track: Upload a photo or video and let Gemini compose a track inspired by its visual content
Availability and Verification
Lyria 3 is available in the Gemini app for users 18+ across multiple languages, with higher generation limits for paid subscribers. The feature is also rolling out to YouTube creators through Dream Track.
All generated music is embedded with SynthID, Google's imperceptible watermark for AI-generated content. Users can verify whether audio was created by Google AI by uploading files directly to Gemini, which checks for SynthID and provides verification feedback.