- Google has introduced AI music creation to the Gemini app using DeepMind’s new Lyria 3 model, now rolling out in beta.
- Users can generate 30-second songs from text prompts, photos, or videos, with automatic lyrics and custom cover art.
- Lyria 3 is also integrated into YouTube Dream Track to help creators produce custom soundtracks for Shorts.
Google has introduced AI-powered music generation to the Gemini app to expand its suite of creative tools beyond images and video. The new feature uses Lyria 3, the latest generative music model from Google DeepMind, and is now rolling out in beta. With this update, users can create short, custom songs directly in the Gemini app using simple text prompts or by uploading photos and videos.
How Lyria 3 works in the Gemini app
Users can describe the type of song they want, including the genre, mood, or theme. For example, someone can ask Gemini to create “a fun afrobeat song about family memories” or “a comical R&B slow jam about a sock finding its match.” Within seconds, the app generates a 30-second track.
Lyria 3 offers several improvements over earlier versions:
- It automatically generates lyrics based on the prompt.
- It gives users more control over style, vocals, and tempo.
- It produces more realistic and musically detailed tracks.
Users can also upload a photo or video, and Gemini will create a soundtrack that matches the content. Each track comes with custom cover art generated by Google’s Nano Banana model. Songs can be downloaded or shared through a link.
Google says the goal is to help people express themselves in a fun and creative way, rather than to produce professional music releases.
Integration with YouTube Shorts
Lyria 3 is also being integrated into YouTube’s Dream Track feature. Available in the U.S. and expanding to more countries, the tool helps creators generate custom soundtracks for YouTube Shorts.
According to Google, the updated model improves audio quality and allows creators to better customize music for their videos.
AI watermarking and verification
All music created in the Gemini app includes SynthID, Google’s invisible watermark for identifying AI-generated content.
Gemini’s verification tools now support audio, in addition to images and videos. Users can upload a file and ask whether it was generated using Google AI. The system checks for SynthID and analyzes the file before responding.
Responsible AI development and copyright safeguards
Google said it developed Lyria 3 with input from the music community and took copyright and licensing issues into account during training.
The system is designed for original music creation, not for copying specific artists. If users mention an artist’s name, Gemini treats it as general inspiration rather than imitation. Google also uses filters to compare generated tracks against existing content and allows users to report potential copyright issues.
Availability and subscription tiers
Lyria 3 is available in beta within the Gemini app for users aged 18 and older. At launch, it supports English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese. The feature is rolling out on desktop right away, with mobile access expanding over the next few days.
Google AI Plus, Pro, and Ultra subscribers will receive higher usage limits compared to standard users.
