On February 18, 2026, Google DeepMind officially announced that its latest music‑generation model Lyria 3 is now available in the Gemini App. The announcement quickly became a focal point in global tech and music circles. Lyria 3 is Google’s most advanced music‑generation model to date: users can generate a 30‑second high‑quality custom track with a single click by entering text or uploading images or videos; it can automatically create lyrics, vocal performances and accompaniment, and allows fine control over style, tempo and emotion.
This integration marks the move of DeepMind’s music AI capabilities from the lab to an audience of hundreds of millions: Gemini App’s monthly active users have surpassed 750 million. Lyria 3 is no longer a pure research tool but a "music collaboration partner" accessible to casual users, creators and professional musicians. At the same time, YouTube’s Dream Track feature has been opened to creators worldwide, further expanding Lyria 3’s reach.
Core technical breakthroughs of Lyria 3: from “music generation” to “collaborative composition”
Lyria 3 represents a comprehensive upgrade over previous Lyria models. Google says it delivers significant improvements in realism, musical complexity and naturalness, capable of producing professional‑quality full tracks with believable vocal performances, rich instrumental layers and natural transitions.
Main feature highlights:
- Multimodal input: Supports pure text prompts (e.g., “a relaxed summer jazz piece with a soft female vocal”) or uploading images/videos as references; the AI will automatically match mood and style to visual content.
- Complete music generation: Automatically generates lyrics, vocal performance, accompaniment and mixing; users can obtain a finished piece without providing lyrics.
- Fine‑grained control: Lets users specify style, rhythm, tempo, emotional intensity and even vocal type (male/female/chorus), with iterative editing supported.
- 30‑second limit: To mitigate misuse and copyright issues, single‑generation output is currently capped at 30 seconds, with support for seamless stitching.
- SynthID watermark: All generated content contains an invisible watermark to assist tracking and copyright protection.
Compared with earlier Lyria versions, Lyria 3 achieves qualitative leaps in musical coherence, harmonic complexity and vocal naturalness. Google DeepMind emphasizes that Lyria 3 is not simply imitating existing music but is designed to “help users explore new ideas and create the music they want to hear.”
Technically, Lyria 3 builds on DeepMind’s long‑term work in audio, combining diffusion models and advanced audio encoding techniques to achieve high‑fidelity output. Training data were strictly curated, and the model incorporates safeguards against imitating artists’ styles too closely, helping avoid generation that is highly similar to existing performers’ works.
Availability and global rollout: rapid coverage from desktop to mobile
Lyria 3 is now live on the Gemini web app, with mobile rollout to follow in the coming days. It is available to users aged 18 and older worldwide and supports eight languages at launch: English, Japanese, Korean, Hindi, Spanish, Portuguese, German and French, with more to be added over time.
YouTube creators can also access Lyria 3 via the Dream Track feature; previously limited to U.S. users, this capability has now expanded to many more regions globally. Users can select the “Create Music” tool in the Gemini App or visit gemini.google.com/music to try it.
Currently in beta, Google stresses this is an "experimental feature" intended to let users express creativity in a fun, safe way, not to replace professional music production.
Deep impact on the music industry: opportunities and challenges
The deployment of Lyria 3 is seen by the industry as a key step in moving AI music generation from a “toy” to a “production” level. It greatly lowers the barrier to music creation: casual users can generate professional‑sounding short pieces in seconds, independent musicians can quickly validate ideas, and advertisers and short‑form video creators can significantly reduce soundtrack costs.
For the professional music industry, this represents both opportunity and challenge. The opportunity lies in AI as an assistive tool, helping composers generate demos, expand inspiration and handle repetitive accompaniment tasks, freeing human creators to focus on emotional expression and innovation. Many musicians have begun experimenting by using Lyria 3 to generate material and then adapting or performing it with live musicians, creating new “human–machine co‑creation” workflows.
The challenges are clear as well. The industry relies on copyright protection and artists’ uniqueness; widespread AI‑generated content may intensify infringement disputes and market saturation. Google has implemented SynthID watermarking and usage limits as mitigations, but balancing innovation with copyright protection remains a global regulatory question.
A deeper question concerns the “soul” of music. Critics worry that overreliance on AI could strip music of uniquely human emotional nuance and lead to homogenized content. Supporters counter that AI is a tool and that final quality still depends on creators’ aesthetics and storytelling. Just as photography did not eliminate painting but spawned new movements, AI music could give rise to new creative paradigms.
Conclusion: the beginning of the AI music era
Google Gemini’s integration of Lyria 3 means AI music generation has officially moved from the lab into everyday life. It makes “composing” no longer the exclusive domain of professionals but a creative activity anyone can join.
The emergence of Lyria 3 is both a disruption to the traditional music industry and a new empowerment of human creativity. In the future, AI and human musicians are likely to form a closer symbiosis—AI handling efficient generation while humans provide emotional depth and final judgment. This may be the right way to approach music in the AI era.
In a year of rapid technical progress, Lyria 3 is only the beginning. The next milestones may include AI‑generated full albums, real‑time interactive concerts, or AI‑original music works with genuine emotional depth.
Music has never been this accessible.
▶️ Enjoy this article? Creating is hard!
Follow our YouTube channel:
AI Rosetta
Your likes, subscriptions and shares are our greatest support! 🙏
More AI news at: https://www.airosetta.com