- What: Google and Google DeepMind released Lyria 3 Pro, an advanced music generation model.
- Key Upgrade: Generates tracks up to 3 minutes long, a 6x increase over the previous 30-second limit.
- Availability: Rolling out to Gemini paid subscribers, Google Vids, ProducerAI, Vertex AI, and the Gemini API.
- Safety: Every track is embedded with SynthID watermarking to identify AI-generated content.
Google and Google DeepMind have officially launched Lyria 3 Pro, a sophisticated music generation model that shatters previous technical boundaries by producing full-length tracks up to three minutes long. Released just one month after the debut of its predecessor, the Pro model marks a massive 600% increase in track length, transitioning the technology from short social clips to complete musical compositions.
Better Control and Track Structure
The release of Lyria 3 Pro represents a significant leap in how AI understands the architecture of music. While Lyria 3 was primarily used for short, 30-second snippets, the Pro model is designed for comprehensive creative control. According to Google, the model now understands complex track structures, allowing users to define specific elements such as intros, verses, choruses, and bridges within their prompts.
This structural awareness addresses a major pain point in generative audio: the tendency for AI to produce aimless, "drifting" melodies. By allowing users to specify the arc of a 180-second piece, Google is positioning Lyria 3 Pro as a tool for professional-grade drafting and creative experimentation rather than just a consumer novelty.
Broad Ecosystem Integration
Google is aggressively deploying Lyria 3 Pro across its entire software stack. The rollout includes:
- Gemini App: Paid subscribers (Google AI Pro and Ultra) gain immediate access to longer generations.
- Google Vids: The video editing app will use the model to allow users to generate custom soundtracks for their projects.
- ProducerAI: The generative AI-powered music production tool, which Google acquired last month, will be powered by the new model.
- Developer Tools: Lyria 3 Pro is now in public preview on Vertex AI and is accessible via the Gemini API and AI Studio, enabling enterprise developers to build custom music generation workflows.
Training Ethics and the "Mimicry" Debate
The question of training data remains a flashpoint in the AI music industry. Google emphasized that Lyria 3 Pro was trained using data from its partners as well as "permissible data" harvested from YouTube and Google’s own repositories.
Addressing the concerns of the music industry and copyright holders, Google stated that the model is specifically designed not to "mimic" the unique voice or style of a specific artist. However, the company noted that if a user includes a specific artist’s name in a prompt, the model will take "broad inspiration" from that artist to inform the generated track’s vibe or genre.
To ensure transparency, all tracks generated by both Lyria 3 and Lyria 3 Pro are marked with SynthID. This imperceptible watermarking technology allows platforms and users to detect whether a piece of music was created or edited using AI, providing a layer of protection against the proliferation of "AI slop" or misattributed content.
Competitive Landscape and Industry Impact
The launch of Lyria 3 Pro comes during a week of heightened activity in the digital music space. Earlier this week, Spotify introduced new tools to allow artists to review songs released under their names, a move designed to prevent bad actors from flooding the platform with AI-generated tracks misattributed to famous performers. Similarly, the streaming service Deezer has launched tools to help identify AI-generated music across the industry.
Google’s decision to bake SynthID directly into the Lyria output suggests an attempt to set an industry standard for responsible generation while simultaneously competing with other music AI startups. By offering a 3-minute generation window, Google has effectively moved the goalposts for what users expect from text-to-music technology.
Impact for Developers and Creators
For developers, the inclusion of Lyria 3 Pro in Vertex AI and the Gemini API opens the door for a new class of creative applications. The ability to programmatically generate structured, 3-minute musical pieces could revolutionize background music for gaming, automated podcast scoring, and personalized advertising.
For creators, the integration into Google Vids and Gemini Pro means that "high-fidelity" music production is becoming increasingly democratized.
"This changes how developers will build the next generation of creative apps, moving AI music from a 30-second novelty to a legitimate production-ready format," according to industry analysts.
What's Next
As Lyria 3 Pro rolls out to paid subscribers and enterprise users, the focus will likely shift to the quality of the "broad inspiration" it takes from artists and whether the 3-minute limit can be extended even further. With Vertex AI currently in public preview, third-party integrations are expected to surge in the coming months, potentially leading to a new era of AI-integrated music production tools.