If you’ve ever found yourself humming a tune that doesn’t exist, or worse, trying to explain a “vibe” to a friend through a series of awkward hand gestures, then you need to check out what Google has just delivered. The ultimate creative lifeline via the Gemini app that integrates Google DeepMind’s most sophisticated music generation model to date: Lyria 3.
Gone are the days when AI was just for generating quirky emails or debatably useful life advice. Now, anyone with a Gemini subscription can transform a simple text prompt or a stray photo into a high-quality, 30-second soundtrack… yes a photo.
From Socks to Studio Quality
What makes Lyria 3 a genuine game-changer isn’t just the audio quality; it’s the sheer creative freedom. Unlike earlier models that required you to be a closeted poet, Lyria 3 generates the lyrics for you. Want a “comical R&B slow jam about a lonely sock finding its match”? Gemini will pen the heartbreak and provide the smooth bassline in seconds.
The update introduces three major improvements that elevate the experience from “novelty” to “noteworthy”:
- Automated Lyrics: No need to stress over rhyming “banana” with “cabana” the AI handles the words based on your prompt.
- Finer Control: You can now tweak the style, vocals, and tempo to ensure the track actually matches the mood you’re after.
- Complex Compositions: The tracks feel more realistic and musically layered, moving away from the “elevator music” vibes of yesteryear.
How to Start Your Music Career (From Your Couch)
The interface is impressively intuitive. You can go the traditional route with Text-to-Track, asking for something specific like a “fun Afrobeat track with a true African vibe” to celebrate your mum’s home-cooked plantains.
But the real magic lies in the Photo-to-Track feature. If you’ve just come back from a weekend hike, you can upload a photo of your dog, Spot, exploring the woods. Gemini will analyse the visuals and compose a track with lyrics that perfectly capture Spot’s outdoor adventures. It’s custom, it’s catchy, and it comes with unique cover art generated by Nano Banana to make it “share-ready” instantly.
Responsible Riffs
Of course, with great power comes the inevitable question: “Is this going to replace my favourite artist?” Google is quick to clarify that Lyria 3 is built for original expression, not for mimicking existing stars. If you ask for a track “in the style of” a famous singer, Gemini uses that as broad inspiration rather than a blueprint for a deepfake. it’s “sound-alike” rather than “plagerism”.
To keep things transparent, every track is embedded with SynthID. This is an imperceptible watermark that identifies the audio as AI-generated. You can even upload a file to Gemini and ask, “Was this made by Google AI?” and it will use its own reasoning to verify the track’s origins. A great feature for determining the difference between human made music over AI generated.
What do we think?
While these 30-second clips might not be winning a Grammy anytime soon, that’s not really the point. Lyria 3 is about adding a personal soundtrack to your daily life. Whether you’re a YouTube creator looking for a “vibey” backing track for a Short or just someone who wants to send a custom birthday song to a mate, the barrier to entry for music creation has officially vanished.
Music generation is currently rolling out to Gemini users (18+) in several languages, including English, Spanish, and French. If you’re a Google AI Premium subscriber, you’ll even get higher limits to keep the hits coming.
Head over to gemini.google.com and start prompting.
