Google is putting a professional-grade music studio inside a chat window. The company just pushed Lyria 3.5, its most advanced text-to-music model yet, into both the Gemini app and the Gemini API, meaning anyone with a prompt can now generate full tracks with expressive vocals and layered instrumentation, no DAW or music theory degree required.
Google just gave millions of Gemini users a new toy: a music studio that lives inside a chat prompt. The company announced today that Lyria 3.5, described as its best-sounding music generation model to date, is now live in the Gemini app and available through the Gemini API for developers who want to bake AI-generated soundtracks into their own apps.
According to Google's announcement, Lyria 3.5 brings 'more expressive vocals and richer musical arrangements' than what came before it. That's a notable claim in a space where the gap between AI-generated music and human-made tracks has been closing fast, and where the biggest complaint from users has always been robotic-sounding vocals or thin, repetitive instrumentation.
This isn't Google's first swing at generative audio. Lyria has been quietly powering music features across Google's ecosystem for a while now, but folding the latest version directly into Gemini changes the distribution math entirely. Instead of a standalone tool musicians and hobbyists have to seek out, Lyria 3.5 becomes something hundreds of millions of Gemini users can stumble into mid-conversation, just by asking for a song.
The timing matters too. Generative music has become one of the more contested corners of the AI boom over the past two years, with startups like Suno and Udio racking up huge user numbers and, in Udio's case, a licensing deal with Universal Music Group after a lawsuit. Google entering this space with Gemini's massive built-in audience puts real pressure on those startups, who don't have anything close to Google's distribution or infrastructure.
Google hasn't published detailed benchmarks comparing Lyria 3.5 to competitors, and the blog post is light on technical specifics, focusing instead on the vocal and arrangement improvements. That's typical for these consumer-facing rollouts. The real test will come from users pushing prompts through the model and comparing output quality against Suno's latest versions or OpenAI's own audio experiments.
What's notably absent from the announcement is any detailed discussion of training data or how Google is handling copyright concerns, an issue that's dogged nearly every generative music player in this space. Labels have been aggressive about protecting their catalogs, and any AI company shipping a mainstream music tool without addressing provenance risks running into the same legal headaches Suno and Udio have already faced.
For now, Lyria 3.5 is rolling out as a feature update rather than a standalone product launch, which fits Google's broader playbook of embedding AI capabilities directly into Gemini rather than spinning up separate apps. It's the same strategy the company has used with image generation and video tools, folding new modalities into one interface so users don't have to leave the app to create.
The bigger question is what this means for the music industry longer term. As tools like Lyria 3.5 make full-track generation as easy as typing a sentence, the barrier to producing passable music drops to nearly zero. That's exciting for casual creators and content producers who need quick background tracks, but it also raises the stakes for professional musicians and labels watching AI-generated content flood platforms like YouTube and Spotify.
Google didn't say whether Lyria 3.5 output will carry any watermarking or labeling to distinguish it from human-made tracks, a detail that's become increasingly important as platforms grapple with disclosure requirements for AI content. That's likely the next shoe to drop as scrutiny around AI-generated media keeps intensifying across every format, not just music.
Lyria 3.5's arrival inside Gemini signals Google is done treating AI music as a side experiment and is ready to compete head-on with dedicated players like Suno and Udio, backed by an audience those startups can only dream of reaching. For casual users, it means better, more expressive AI-generated tracks are now just a prompt away. For the music industry, it's another sign that the line between human and machine-made music is getting blurrier by the month, and the conversations around credit, copyright, and disclosure aren't going away anytime soon.
