Gemini can now also create music—welcome to Lyria 3

Artificial intelligence expands its creative horizons further. Starting today, Gemini enriches its arsenal by introducing the generation of personalized music tracks, based on simple textual or visual cues.

This new feature is based on Lyria 3, the latest and most advanced sound-generation model developed by Google DeepMind.

Lyria 3 is Google’s AI capable of creating music

Users can generate audio tracks lasting 30 seconds simply by providing an idea, describing a mood, a specific memory, or even by uploading a photograph or a short video. The system translates these diverse inputs in moments into high-quality musical compositions.

One of the most interesting innovations concerns lyrics writing: users of the platform are no longer required to write them themselves, as the algorithm handles it in total autonomy, adapting the words to the requested mood.

The level of creative control offered allows you to define with precision the target genre, the desired vocal style and the tempo of the track.

To complete the multimedia experience, each track is accompanied by a personalized cover and instantly generated by the model Nano Banana, facilitating quick sharing on social channels.

As stated by the Mountain View company, the primary aim of this feature is not to produce complex academic masterpieces, but to offer a fun and personal expressive medium to communicate with one’s circle of friends.

Global distribution and access

The beta phase of this tool is currently being rolled out for both the desktop interface and the mobile application. Access is granted to all adult users and initially supports several languages, including English, German, Spanish, French, Hindi, Japanese, Korean, Portuguese and Italian, with expansion plans already anticipated for the near future.

Although the feature is open to anyone meeting the age requirement, Google has introduced specific advantages for subscribers to paid plans. AI Plus, Pro and Ultra tier subscribers will enjoy significantly higher usage limits than free-tier users.

The capabilities of this new tool will not be limited to the main application alone. Creators on YouTube will be able to leverage this same technology through the Dream Track program to uniquely customize the soundtrack of their Shorts, making the audio accompaniment increasingly aligned with the visuals in the video.

Transparency and copyright respect

In an industry where protecting intellectual property represents a central and highly complex challenge, Google has integrated structural security measures.

Each produced composition incorporates SynthID, a digital watermark imperceptible to the human ear but essential for unequivocal identification of content generated by artificial intelligence.

Additionally there is a new verification tool directly integrated into the interface. By uploading an audio file, the system will be able to analyze it and confirm or deny the presence of the SynthID mark.

To prevent plagiarism and commercial abuses, the infrastructure has been rigorously designed to promote only original expression.

If the user enters the name of a real artist or famous band within the text prompt, the algorithm will interpret it solely as a vague stylistic inspiration, avoiding direct imitation of the voice or distinctive sonic identity.

There are also strictly active cross-check filters that compare results with existing copyrighted works.

The company has also set up dedicated channels for the rapid reporting of potential rights violations, ensuring a measured and compliant approach with current intellectual property regulations.