Google puts Lyria 3.5 music AI inside the Gemini app
The model writes vocals and arrangements on request and also runs in Flow Music, AI Studio and Vids. Google says training data was licensed.
Symbolic image: in a small studio a hand pushes a fader on the mixing console while a phone in a stand shows abstract waveform blocks.
Google has switched on Lyria 3.5, so people can now generate tracks with vocals directly in the Gemini app or through an API.
At a glance
- Lyria 3.5 is available in the Gemini app and through an API.
- Google claims more expressive vocals and richer arrangements than the previous version.
- Other entry points: Google Flow Music, Google AI Studio and Google Vids.
- Google says only licensed content was used, but gives no training-data specifics.
Google has turned on Lyria 3.5 inside the Gemini app and released it through an API. Type a description and the model hands back a finished track, either sung or purely instrumental. Google says this version delivers more expressive vocals and richer musical arrangements than its predecessor.
What the Gemini app offers
Users pick a genre and a style, choose between vocal and instrumental output, and decide whether they want something short or longer. Templates cover the obvious starting points: background music for a video, or a personalized birthday song. The framing points at casual users rather than working musicians.
Four ways into the same model
Beyond Gemini, Lyria 3.5 runs in Google Flow Music with extended controls, in Google AI Studio for developers, and in Google Vids. The API opens the model to outside products. One model now reaches a consumer chat app, a creative tool and a developer platform at the same time.
The licensing claim
Google states that the Lyria model was trained exclusively on licensed content, and uses that to set itself apart from Suno. It offers no specifics on which catalogs or datasets went into training. That leaves the claim as a company statement no outside party can verify.
What stays open
The available reporting does not say which countries and languages Lyria 3.5 covers, what limits apply to track length or volume, or what the API costs. Nothing is stated about the rights users hold in generated songs, or about how generated audio is labeled. Those points are unverified here rather than rounded into claims.
FAQ
What is Lyria 3.5?
It is Google's music generation model, which turns a text description into a track with or without vocals. It succeeds the earlier Lyria version and is pitched on more expressive singing.
Where can I use Lyria 3.5?
In the Gemini app, in Google Flow Music with extended features, in Google AI Studio, in Google Vids, and through an API for your own applications.
Was Lyria 3.5 trained on licensed music?
Google says yes and presents that as the difference from Suno. It does not name the catalogs or datasets involved, so the claim cannot be checked from outside.
