Google has announced that it is adding a music-generation feature to the Gemini app, powered by Google DeepMind’s new Lyria 3 model. The feature, currently in beta, allows users to create original songs using simple text prompts.
To generate music, users describe the type of track they want, and Gemini produces a 30-second song complete with lyrics and AI-generated cover art. For example, a prompt such as “a comical R&B slow jam about a sock finding its match” will result in a fully composed track tailored to that idea. Users can also upload a photo or video, and the tool will compose a song that matches the mood of the media.
According to Google, Lyria 3 improves on previous models by producing more realistic and complex compositions. Users can fine-tune elements such as style, vocals and tempo to better shape the output.
Alongside its integration into Gemini, Lyria 3 is also being rolled out to creators on YouTube through the Dream Track feature. Previously limited to U.S.-based creators, Dream Track will now be available globally, expanding access to AI-powered music tools.
Google stressed that the system is designed for original expression rather than imitation. While users can reference an artist’s name in a prompt, Gemini will only generate music inspired by a similar style or mood, not replicate a specific artist. The company added that safeguards and filtering systems are in place to prevent outputs from infringing on existing works.
All tracks generated using Lyria 3 will include a SynthID watermark to clearly identify them as AI-created. Gemini will also gain tools to detect AI-generated music, allowing users to upload tracks and check whether they were produced using AI.
The music-generation feature is rolling out to Gemini users aged 18 and over worldwide, with support for English, German, Spanish, French, Hindi, Japanese, Korean and Portuguese.
