What you need to know
- Gemini now generates 30-second songs with lyrics from text prompts or photos.
- Google’s Lyria 3 model powers the feature, handling lyrics, style control, and more realistic audio.
- Type a prompt or upload an image/video, and Gemini turns it into a share-ready song with custom cover art.
The Gemini app has moved beyond just text and images. As of today, it wants to write your next song.
Gemini has launched music generation in beta, using Lyria 3, Google DeepMind’s newest generative music model. This means you can now create a 30-second track with lyrics in seconds, just by giving a prompt or even a photo.
Google says Lyria 3 improves on earlier versions in three main ways: it can generate lyrics automatically, lets you control style, vocals, and tempo more closely, and creates tracks that sound more realistic and layered. In simple terms, you don’t have to be a songwriter. Just describe a vibe, like “a nostalgic afrobeat tribute to my mom’s cooking” or “a goofy R&B jam about a lonely sock,” and you’ll get a polished mini-track in return.
Photo or video to track
You can also upload a photo or video, and Gemini will turn that moment into a soundtrack with matching lyrics. For example, a hiking photo of your dog can become a woodland anthem with lyrics. The system even creates custom cover art using Nano Banana, so your track is ready to share right from the app.
Since its launch, Gemini has added more creative tools, moving from images to video. Music is the next step. However, Google isn’t trying to make this a Spotify replacement or a hit-making machine. The aim is to create a fun way to add music to your daily life without using complex software or hiring a producer.
There’s also something for creators. Lyria 3 is coming to Dream Track on YouTube Shorts, starting in the U.S. and rolling out worldwide. This means you can get better, customizable backing tracks and lyrics for short videos, which is important since audio can really affect engagement.
Of course, AI music does bring up copyright concerns. Google says Lyria 3 is trained with careful attention to copyright and partner agreements, and it’s meant for original expression, not copying. If you mention a specific artist, Gemini uses it as inspiration, not as a direct guide. There are filters to check for existing content, and users can report possible violations.
Each generated track includes SynthID, Google’s invisible watermark for identifying AI-generated content. Gemini’s verification tools now work with audio too, so you can upload a file and ask if it was made with Google AI. The system checks for SynthID and uses its own analysis before answering.
Lyria 3 is now available in the Gemini app for users 18 and older, in several languages like English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese. It’s launching on desktop first, with mobile coming soon. Google AI Plus, Pro, and Ultra subscribers have higher usage limits.
Android Central’s Take
Personally, I think this is one of the more practical AI upgrades we’ve seen. Not because it’ll replace musicians — it won’t — but because it lowers the barrier to creative expression in a way that feels playful instead of intimidating. If you’ve ever wanted a custom birthday song, a theme for your group chat, or a quirky track to post on Shorts, you can now get one in seconds. For users, that means creativity on demand, and that’s a pretty compelling direction for AI to take.
