Google's Lyria 3.5 AI Music Model Launches

AI music model Lyria 3.5 launched globally for all users on the web and began rolling out inside Gemini on September 4, Google (GOOGL) said, extending the company’s push into machine-generated audio production.

Key Takeaways

Lyria 3.5 launched globally for web users and began rolling out inside Gemini on September 4
The model also extends into Google Flow, the company’s AI video and audio creation tool
Lyria began as a research project in 2023, first reaching the public through YouTube’s Dream Track feature
Suno and Udio have already drawn lawsuits from major record labels over training data

What The Lyria 3.5 Rollout Actually Includes

The launch also extends into Google Flow, the company’s AI video and audio creation tool, according to a post from Google’s official account. Users type a text prompt describing mood, instrumentation or genre, and Lyria generates an original track rather than remixing existing songs.

Google pointed listeners to a tool for testing the update directly in the browser.

A wider rollout to the Gemini app is expected within days of the September 4 post.

How Lyria Actually Works

Lyria is Google DeepMind’s text-to-music system, a neural network trained on audio data to translate written descriptions into original compositions instead of splicing samples together. It outputs a finished waveform, meaning sound emerges directly from the model’s own calculations rather than from prerecorded loops.

Also Read: Multiverse’s Quasar 438B is First European Model to Top AI Index

From YouTube Experiment To Gemini Feature

Lyria began as a research project Google DeepMind introduced in 2023, first reaching the public through YouTube’s Dream Track feature, which let a small pool of creators generate short backing tracks patterned after licensed artists.

A second version followed in 2024 with wider access through Google’s Vertex AI platform for enterprise customers.

Version 3.5 marks the first time the AI music model has shipped as a consumer feature inside Gemini, rather than through a separate experimental app.

The Competitive Stakes Behind Consumer AI Music

Releases from Suno and Udio have already drawn lawsuits from major record labels over training data. Google’s broader rollout raises the same questions at a larger scale.

Universal Music Group and other labels have pushed for licensing deals rather than outright bans on AI-generated tracks.

How Google handles artist consent and revenue sharing for Lyria 3.5 will determine whether labels treat Gemini’s new feature as a partner or a target.