Lyria 3 Pro

What is Lyria 3 Pro?

Lyria 3 Pro is a music generation model designed to produce longer tracks with structural awareness. It’s positioned as an advanced version of Lyria 3, intended for people and organizations that want more control over how a song unfolds across sections.

In addition to standalone use, Lyria 3 Pro is being integrated into more Google products where professionals and creators already work—such as Vertex AI, Google AI Studio, Vids, and the Gemini app—so users can generate or add music as part of broader creative workflows.

Key Features

• Longer track generation (up to 3 minutes): Creates extended compositions rather than shorter snippets, giving more space for structure.
• Structural awareness for music composition: Better understanding of how musical sections fit together, aimed at producing more coherent arrangements.
• Prompting for specific song elements: You can request elements like intros, verses, choruses, and bridges to guide the output.
• Customization and creative control: Designed to support experimentation and different styles, including tracks that involve complex transitions.
• Integrations across Google products: Lyria 3 Pro is made available through Vertex AI (public preview), Google AI Studio and the Gemini API, and is included in Google Vids and the Gemini app (availability described below).
• Output identification via SynthID watermark: Outputs are embedded with SynthID, an imperceptible watermark for identifying Google AI-generated content.

How to Use Lyria 3 Pro

1. Choose where you want to generate music: Use Lyria 3 Pro via one of the mentioned integrations (Vertex AI, Google AI Studio / Gemini API, Vids, or Gemini app).
2. Write prompts that specify structure: Include requests for sections such as intro, verse, chorus, and bridge when you want a particular layout.
3. Iterate for style and transitions: Try different prompt instructions to steer the output toward the style and flow you want.
4. Use generated music in downstream workflows: Where supported (for example, in Vids), add the generated music to content such as marketing videos or creative projects.

Use Cases

• Gaming and bespoke soundtrack workflows (on-demand at scale): Use Lyria 3 Pro in Vertex AI to rapidly generate original soundtracks for game projects.
• Developer-built creative tools (coherent structure): In Google AI Studio and via the Gemini API, incorporate improved musical awareness and structural coherence to generate content inside custom applications.
• Video creation with matched background music: In Google Vids, generate custom music with Lyria 3 or Lyria 3 Pro and add it to videos to match a creator’s style.
• Longer audio generation for vlogs, podcasts, and tutorials: In the Gemini app, use Lyria 3 Pro to generate longer generations with more room to add details aligned to the content.
• Song iteration with an agentic music workflow (ProducerAI): Use ProducerAI, described as a collaborative music creation tool, together with Lyria 3 Pro to iterate on comprehensive songs.

FAQ

Where is Lyria 3 Pro available?

The page states Lyria 3 Pro is available via Vertex AI (public preview), Google AI Studio and the Gemini API, and it is also rolling out in Google Vids and the Gemini app to certain audiences (paid subscribers for Gemini app; Workspace customers and AI Pro & Ultra subscribers for Vids, as described in the post).

How long can Lyria 3 Pro generate tracks?

Lyria 3 Pro can create tracks up to 3 minutes long.

Can I ask for specific parts of a song?

Yes. The page says you can prompt for specific elements such as intros, verses, choruses, and bridges.

Does the model mark AI-generated outputs?

Yes. All Lyria 3 and Lyria 3 Pro outputs are embedded with SynthID, an imperceptible watermark intended to identify Google AI-generated content.

Are there usage limitations or IP protections mentioned?

The post describes responsibility measures: it says Lyria 3 and Gemini do not mimic artists when a prompt names a creator (treated as broad inspiration), and that outputs are checked with filters against existing content. It also notes users must follow Google’s Terms of Service and Gen AI prohibited use policies, including restrictions related to intellectual property and privacy.

Alternatives

• Other music generation models (standalone or in-app): If you only need short prompts-to-audio generation, other AI music generators can be used, but they may not offer the same emphasis on structural awareness and multi-section control described here.
• General-purpose generative audio tools: Tools that generate audio based on style or instrument prompts may be suitable for experimentation, though they may not support the same section-by-section prompting (intro/verse/chorus/bridge).
• Video tools with built-in music libraries: For workflows focused on quick music selection rather than generative composition, standard music libraries and editors can substitute the need to generate tracks; the trade-off is less custom compositional control.
• Artist-first composition workflows (human composition + editing): For maximum creative direction and precise control, a traditional composition/editing workflow can complement AI generation, especially when you need to match specific references that are difficult to encode in prompts.