Google has expanded its AI ambitions by adding music generation to its Gemini app.
The new beta tool allows users to create short, personalized tracks in seconds — no music production skills required.
The feature runs on Google DeepMind’s Lyria 3 model, marking a major step forward in AI-generated audio.
Users can describe a genre, mood, memory, joke, or concept through a simple text prompt. The system then produces a 30-second music track, complete with instrumental elements or automatically generated lyrics.
Lyria 3 introduces automatic lyric writing, removing the need for users to craft their own words. It also offers greater control over musical style, tempo, and vocals, delivering more complex arrangements than earlier versions.
Turning images and videos
Music generation in Gemini works in two main ways.
Users can either input text prompts or upload photos and videos. The AI interprets visual content and generates a matching soundtrack designed for quick sharing rather than studio-level production.
Each generated track includes custom cover art created by Nano Banana. Users can download the music file or share it through a link, making it easy to add background audio to social posts or personal moments.
Lyria 3 is also being integrated into YouTube Dream Track, enhancing tools available to Shorts creators.
The system allows creators to generate custom background music or vocal segments that align with the tone of short-form videos. The rollout began in the United States and is gradually expanding to additional countries.
Built-in watermarking
Google says all AI-generated audio in the Gemini app includes SynthID, an invisible watermark designed to identify AI-created content.
The app now supports audio verification alongside existing image and video checks. Users can upload files and ask whether they were generated using Google’s AI tools.
Addressing copyright
During development, Google says it worked closely with members of the music community to address copyright and licensing concerns.
Lyria 3 is designed to support original expression rather than replicate specific artists. If prompts include an artist’s name, the system treats it as stylistic inspiration instead of direct imitation.
Filters compare outputs against existing works, and users can report content that may infringe on rights. Access to the feature requires compliance with platform rules covering intellectual property and privacy.
Lyria 3-powered music generation is available in beta within the Gemini app for users aged 18 and older.
The tool supports multiple languages, including English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese.
The rollout is beginning on desktop, with mobile access expanding shortly after. Paid Google AI subscription tiers will receive higher usage limits.
Users can try the feature at gemini.google.com.
