Google announced on Wednesday that it is adding a music-generation feature to the Gemini app. The feature is powered by DeepMind’s Lyria 3 music-generation model and is currently in beta.
Users describe the song they want to create via a text prompt, and the app generates a 30-second track with lyrics and cover art. The cover art is created by Nano Banana. For example, asking for a “comical R&B slow jam about a sock finding its match” results in a generated track. Users can also upload a photo or video, and the tool creates a song matching the mood of the media file.
Lyria 3 offers improvements over previous models, creating more realistic and complex music. Users have control over elements such as style, vocals, and tempo.
Alongside the Gemini rollout, Google is making Lyria 3 available to YouTube creators through the Dream Track feature. Previously, Dream Track was limited to U.S. creators. The release expands availability globally.
Google restricts artist mimicry. If a user names a specific artist in a prompt, Gemini creates a track in a similar style or mood rather than a direct imitation. The company stated that filters check outputs against existing content to ensure original expression.
All songs generated with Lyria 3 carry a SynthID watermark to identify AI content. Gemini is also gaining the ability to detect AI-generated music. Users can upload tracks and ask the app to determine if they are AI-generated.
The feature is rolling out to all Gemini users aged 18 and older worldwide. Supported languages include English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese.
The industry rollout occurs amid ongoing tension regarding AI music. YouTube and Spotify are embracing AI and signing deals with music labels to monetize generated tracks. Meanwhile, AI companies face copyright lawsuits from the music industry regarding training materials. Platforms like Deezer have deployed tools to flag AI-generated music to prevent fraudulent streams.
