Google Just Added AI Music Generation to Gemini. Here Is What It Actually Does.
Google Just Added AI Music Generation to Gemini. Here Is What It Actually Does.
Lyria 3 went live in Gemini today. Google’s most advanced music generation model is now available to all users 18+ — text prompt, image prompt, 30-second track, done.
Here’s what you need to know.
What It Does
You type something like “a comical R&B slow jam about a sock finding their match” and in a few seconds you have a full track — vocals, lyrics, production, the whole thing. You can also upload a photo and it generates music to match the vibe.
Controls include style, vocals, and tempo. Earlier versions made you write your own lyrics. Lyria 3 generates them automatically.
The tracks are 30 seconds long. Make of that what you will.
Cover art is generated automatically via Nano Banana (Google’s image model) and every track gets a SynthID watermark baked in — Google’s invisible fingerprint for AI-generated audio.
Who It’s For
Gemini users who want to make something quickly without any music production knowledge. Content creators who need background tracks without licensing headaches. People who are genuinely curious and want to mess around.
It’s not for producers who know what they’re doing. 30-second tracks with AI lyrics aren’t replacing sessions. They’re replacing the “I just need something for my Instagram story” moment.
The Catch
The artist protection is minimal in practice. Google says if you prompt with a specific artist’s name, it “takes creative inspiration” rather than mimicking them. That’s doing a lot of work. “Sounds like Kendrick but sadder” is a pretty thin guardrail.
SynthID watermarks are imperceptible to humans and don’t prevent you from using the audio — they just exist so Google (and researchers, eventually) can identify it later. This is better than nothing. It’s not a solution to anything.
The real ceiling is 30 seconds. That’s a deliberate limit. They’re not trying to let you make full songs — they’re letting you make clips. Whether that expands depends on how the creative industry responds, and the industry is currently very loudly not okay with any of this.
Availability
Rolling out now in beta. Available in English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese. You need to be 18+.
It’s in the Gemini app — not a separate product, not an API you have to configure. Just there.
Bigger Picture
Lyria 3 is one of three AI music generation tools that actually exist now (Suno and Udio being the others with real user bases). Google has the distribution advantage — Gemini already has users who will just find this feature one day and start using it.
What’s interesting is they’re shipping this as part of a general assistant, not as a dedicated music tool. The bet is that the people who would never seek out “AI music generator” will stumble into it through Gemini. That’s probably correct.
The creative industry’s concern isn’t “AI music generators exist.” It’s “everyone who makes content at scale now has free music generation baked into the tool they already use for everything else.” Different problem.
Lyria 3 is available in the Gemini app starting today.