AI-powered music generation enters a new phase. After conquering image and video, Google takes a logical step by directly integrating a music creation engine into the heart of its Gemini assistant. The model behind all of this is called Lyria 3, developed by Google DeepMind, and it has been available in beta since February 17, 2026. This launch doesn't seem like a discreet experiment: it's a global rollout, for all users aged 18 and over, in eight languages including French, Spanish, and Japanese.
A clear positioning in an already crowded market
The generative music market has existed for several years. Suno and Udio have carved out a solid niche, each with its strengths: Suno for speed and generation of complete songs with lyrics, Udio for audio fidelity and instrumental quality. But these tools remain dedicated platforms, separate from the digital daily lives of most users.
Google is taking a different approach. Rather than offering a competing service to Suno or Udio, the Mountain View firm integrates music creation directly into Gemini, where over 100 million people already ask their questions, create images, or summarize documents. The objective is explicit: to transform Gemini into a complete creative space, without the need to switch applications.
What Lyria 3 can do, concretely
The operation is simple, by design. The user describes an idea in text, for example, 'a funny R&B slow jam about a sock looking for its pair,' and Gemini generates a 30-second track in seconds, including lyrics if desired. Several input modes are available.
Text to audio track : describe a genre, a mood, a personal memory, or a joke among friends to get a song with or without lyrics
Image or video to audio track : upload a visual media and let Gemini compose a track whose lyrics and mood match the content
Creative control : adjust the style, tempo, voices, and musical register according to your preferences
Lyria 3 also marks a break from previous versions by generating lyrics automatically from the prompt, without the user needing to provide one. Each title is accompanied by an album cover generated by the Nano Banana image model.
YouTube and Dream Track: Extension for Creators
Beyond consumer use, Google is extending the deployment of Lyria 3 to video content creators. Dream Track, the musical accompaniment tool for YouTube Shorts, now integrates the model and opens up to creators outside the United States. Previously limited to the American market, this expanded access represents a concrete opportunity for French-speaking and international videographers seeking personalized soundtracks without copyright risks.
YouTube had also launched "Music Assistant" this year in its Creator Music marketplace, a tool allowing members of the YouTube Partner Program to create royalty-free instrumentals via text prompts. Lyria 3 thus fits into a coherent platform strategy rather than an isolated announcement.
The Copyright Question, Addressed Head-On
This is the contentious subject in the music industry, and Google knows it. Suno is facing federal lawsuits in the United States for reproducing sounds from artists like Mariah Carey or Chuck Berry. Google wanted to distinguish itself from the outset on this point.
The firm states that Lyria 3 is "designed for original expression, not to imitate existing artists" . If a user mentions a musician's name in their prompt, Gemini will interpret it as stylistic inspiration, not as an instruction to copy. Filters are in place to control outputs and compare them to existing works. Google itself acknowledges that this approach "is not infallible" and invites rights holders to report any problematic content.
SynthID: Every Title Carries Its Invisible Signature
All titles generated in Gemini integrate SynthID, Google's imperceptible watermark designed to identify content produced by its AI models. This is not new for images and videos, but its extension to audio is a first that enhances traceability.
Gemini also has new audio verification capabilities : the user can upload any audio file and ask Gemini if it was generated by Google AI. The assistant analyzes the present SynthID and uses its own reasoning to provide an answer. An initiative that takes on particular significance in a context where audio disinformation is rapidly developing.
What This Reveals About Google's Strategy
Lyria 3 is not just a creative gadget. It's a demonstration that Google is betting on Gemini as a universal interface layer, capable of processing all forms of human expression in one place. Text, image, video, music: the platform aims to cover the entire creative spectrum.
The fact that Lyria 3 is the first version of Lyria to receive a wide public deployment is significant in itself. Previous versions remained confined to experiments like the Music AI Sandbox or artist partnerships. This time, Google is playing the mass-market card, with a tool that is deliberately accessible, with no learning curve, and available where users already are.
No comments! Be the first one.