Google DeepMind has officially released its latest generation AI music generation model, Lyria 3. The model is now available as a beta version integrated into the Gemini app and is free for users worldwide who are 18 years or older.

The most remarkable feature of Lyria 3 is its full-scenario creative ability. Even users with no musical background can easily generate music through three methods. Users just need to input natural language prompts, such as "a cheerful reggae song suitable for a beach party" or "an epic electronic music about space exploration," and the model can quickly generate a complete 30-second song that includes vocals, lyrics, and instrumental accompaniment, automatically matching the described style, mood, and rhythm.

Google, google

In addition, users can upload photos or screenshots, whether landscapes, portraits, or illustrations, and Lyria 3 will analyze the atmosphere of the image and generate original music that fits the scene, as well as automatically create an album cover. For short video creators, uploading a short video clip allows the model to create a custom background music for the video, perfectly matching the rhythm and emotional tone of the visuals.

However, Lyria 3 currently has two major limitations. First, the generated music has a fixed length of 30 seconds and cannot meet the needs of longer music creation. Second, the model is limited to non-commercial use; for commercial music creation, users need to find other solutions. Google stated that in the future, it will gradually expand the functions, including supporting longer music generation, more language support, and offering commercial licensing options.