Lyria 3 is a generative AI music model that makes short custom tracks, around 30 seconds long. It builds melody, rhythm, instruments, vocals and lyrics from simple prompts. It’s rolling out inside the Gemini app as a beta tool so users can create music from text, photos or audio.
It lets users ask for music in plain words. For example a happy lo fi beat about morning coffee. Or they can upload a photo or short clip and the model makes a track that fits the mood or scene. The result runs about 30 seconds. And it can include original lyrics plus matching cover art made with the audio.
Main features. Automatic lyrics. It writes the words so users don’t have to. Multi input prompts. Text images or video can guide the sound. Style control. Users pick genre mood tempo and vocal tone. Built in watermark. It adds SynthID marks so platforms can spot AI made audio.
Inside the Gemini app users tap Create Music and describe what they want. In a few seconds they get a custom track. They can download it or share it for social posts short reels or small personal projects.
Users say they’ve made all kinds of tracks. An Afrobeat style hype loop. A punk rock type jingle about daily chores. So the tool shows a studio-like range in what it can produce.
If you'd like to access this model, you can explore the following possibilities:
