Gemini Now Lets You Generate AI Music for Free
Google Gemini can help you write text, generate images and video, and write code. Now, the AI bot can generate music too, taking on the likes of Suno when it comes to producing tunes from a simple text prompt. The update is courtesy of the new Lyria 3 audio generation model, which is built into Gemini as of today. Developed by Google DeepMind, Lyria has been accessible in other Google products (such as Vertex AI and YouTube Shorts) to a select number of users, but this is the first time Google is making the model widely available to anyone who wants to try it.
Lyria works the same way as creating images or video: just describe what you want, and the AI does the rest. You might want to hear "a comical R&B slow jam about a sock finding their match" (as per Google's own example), or perhaps "a sea shanty about the dangers of AI slop"—it's up to you.
When you click the "Create music" button inside the Gemini app, you also have the option to pick an existing track to remix, rather than starting from scratch—If you're perhaps stuck for inspiration. These presets cover everything from folk ballads to Latin pop, so you can see the kind of musical scope covered. You can also supply Lyria 3 inside Gemini with an image or video, and get it to compose something that matches the mood of the content you've supplied, including both music and lyrics. The example Google gives is supplying Gemini with a few photos of your dog, and then having it come up with a tune about the pooch and their adventures.
The tracks are limited to 30 seconds each at the moment, and while music making is available to all Gemini users, if you're paying for the Plus, Pro, or Ultra subscriptions, you'll get higher usage limits (though it isn't specified what these are). As per the official announcement blog post, the aim "isn't to create a musical masterpiece, but rather to give you a fun, unique way to express yourself." You're not going to be able to set up your own AI-generated band on Spotify with this, but you can churn out a few entertaining tracks for your own (or someone else's) amusement.
I haven't been able to try out the feature as of yet, but I have heard a few samples that Google supplied. They come across as rather generic and ordinary, exactly as you might expect something to sound that's the averaging out of vast amounts of audio training data—like song genres distilled into their most common ingredients and repackaged.
Google says all created tracks will contain invisible watermarks powered by SynthID, flagging them as AI creations, and you can upload audio tracks to Gemini and run a SynthID check on them. The updated Lyria 3 model is also coming to the Dream Track music maker for YouTube Shorts creators. Lyria 3 is now rolling out inside Gemini, and is available to users aged 18 and above in English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese. It's available first on the web app, and will show up on the mobile app over the next few days. Expansions in "quality and coverage" are planned in the future, Google says.