On Thursday, Google announced that its Gemini app now supports music generation powered by Lyria 3, Google DeepMind’s latest generative audio model.
The feature allows users to create 30-second tracks from simple text prompts or uploaded images, no studio, instruments, or lyrics required.
The update marks Google’s most ambitious push yet into AI-generated music, extending Gemini’s creative toolkit beyond text, images, and video.
From prompts to playback in seconds
“With Lyria 3 embedded in the Gemini app, users can describe a mood, memory, genre, or inside joke and receive a fully produced 30-second track complete with lyrics and instrumentals”, said Joël Yawili, senior product manager, Gemini App and Myriam Hamed Torres, senior product manager, Google DeepMind, wrote in a joint blog post.
A prompt like a comical R&B slow jam about a sock finding their match can generate a catchy track in seconds.
Users can also upload photos or short videos to inspire the soundtrack. For example, a photo dump from a hiking trip could become a custom song about the adventure.
Unlike earlier versions of Google’s music models, Lyria 3 introduces three key upgrades:
- Automatic lyric generation based on prompts
- Greater creative control over style, tempo, and vocals
- More realistic and musically complex outputs
Each track also comes with AI-generated cover art created by Nano Banana, making it easy to share via download or link.
Google says the goal isn’t to replace professional musicians or help users produce chart-topping hits.
Instead, it’s positioning the tool as a lightweight, expressive add-on for everyday creativity, think personalised birthday jingles, inside-joke anthems, or soundtrack snippets for social media.
YouTube creators get an upgrade
Lyria 3 is also powering Dream Track on YouTube, where creators, starting in the U.S. and expanding globally, can generate custom soundtracks for Shorts. The improved model promises higher-quality lyrical verses and backing tracks, potentially reducing reliance on stock audio libraries.
For content creators, this could mean faster turnaround times and more distinctive audio branding.
Built-in AI watermarking
In a nod to growing concerns about AI-generated media, Google says all music produced in Gemini will be embedded with SynthID, its imperceptible watermarking technology for identifying AI-generated content.
The Gemini app will also now support audio verification. Users can upload a file and ask whether it was generated using Google AI. The system checks for SynthID markers and uses additional reasoning signals before returning a response.
Copyright caution and guardrails
Music generation remains one of the most legally sensitive areas in generative AI, particularly around style mimicry and copyright.
Google says Lyria 3 is designed for “original expression,” not direct imitation of existing artists. If users reference a specific musician, Gemini will interpret it as broad stylistic inspiration rather than attempt replication. The company also says it has implemented output filters and copyright safeguards, and allows users to report potential rights violations.
Lyria was first introduced in 2023, and Google says it has worked with the music community and partners while training the new version, paying attention to copyright agreements and responsible development.
Who can access it?
Music generation with Lyria 3 is rolling out in the Gemini app for users aged 18 and above. It is currently available in English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese, with plans to expand language support.
Subscribers to Google AI Plus, Pro, and Ultra tiers will receive higher usage limits.
With OpenAI, Meta, and others racing to dominate generative media, Google’s move signals that AI-generated music is no longer experimental, it’s becoming a built-in feature of everyday consumer apps.
