New Lyria 3 AI Model From Google Can Generate Three-Minute Long Tracks

Google has launched Lyria 3 Pro, an upgraded model of last month’s Lyria 3. The new version can generate music compositions up to three minutes in length.

How it works:

• This expanded function is available on the pro tier.

• The update is designed to better understand song structure, identifying specific elements such as intros, verses, and choruses.

• The prior version could only convert more interpretive prompts, such as genre, mood, or scenario.

• Songs were also limited to 30 seconds.

• The new feature will be available in Google Vids to allow custom music for videos.

• The Vids integration will begin rolling out this week to Google Workspace and Google AI Pro & Ultra.

• Enterprise customers can access Lyria 3 Pro in public preview via Vertex AI. Developers can use it through Google AI Studio and the Gemini API.

• All music created with Lyria 3 will have SynthID, Google’s watermark for AI-generated content.

Training:

• Google has not provided precise details on how Lyria 3 Pro was trained.

• The company stated that it used “materials that YouTube and Google have a right to use under our terms of service, partner agreements, and applicable law.”

• DJ and producer François K and Grammy-winning producer Yung Spielburg were consulted in the development of Lyria 3.

What they said:

• Google: “To protect original expression, Lyria 3 and Gemini do not mimic artists; if a prompt names a creator, the model takes that as broad inspiration. Additionally, we employ filters to check outputs against existing content, and users must adhere to the Terms of Service and Gen AI prohibited use policies, which prohibit violating others’ intellectual property and privacy rights.”