Google introduces Lyria 3, AI that generates professional music | Keryc
Lyria 3 arrives as a new tool to create music with artificial intelligence that sounds coherent and production-ready. Can you imagine generating a track with verses, choruses and natural-sounding vocals without leaving the browser or from your app? Now it's possible for developers in public preview.
What is Lyria 3
Lyria 3 and Lyria 3 Pro are musical generation models that Google is putting in public preview via the Gemini API and an audio experience in Google AI Studio. They're designed to keep musical awareness and structural coherence: that means compositions that flow, with recognizable sections and realistic voices.
Who is it for? For developers, creators, musicians and product teams who want to integrate AI-generated music into apps, games, advertising and social content.
Versions and performance
Lyria 3 Pro (): a premium model aimed at full songs, with tracks up to roughly three minutes. Ideal for studio-quality output and productions that require advanced musical structure.
lyria-3-pro-preview
Lyria 3 Clip (lyria-3-clip-preview): optimized for speed and volume, generates clips of about 30 seconds. Perfect for rapid prototypes, background loops and social media content.
Both variants support realistic voices with expressive nuances, improved clarity and exploration across languages and genres, from pop to funk and Motown.
Creative controls and multimodal inputs
Lyria 3 brings granular controls that let you steer creation with natural language:
Tempo conditioning: ask for a specific tempo like "fast" or "slow" so the track matches the rhythm you need.
Time-aligned lyrics: you can indicate in the prompt when lyrics start and end within the track, useful if you want verses and choruses at specific moments.
Image-to-music input: besides text, you can upload an image to influence mood, style and atmosphere of the audio. A nighttime city photo? The music can take on dark, ambient tones.
These controls make the AI act more like a collaborator that follows precise directions than a box of surprises.
Google AI Studio and access for developers
To experiment right away, Google launches an AI Studio experience. Using a paid API key, you get a dedicated space where you can try Lyria 3 in a playground:
Text mode: describe the music you want with natural language and parameters like tempo or key.
Composer mode: build the song section by section, from the intro to verses and bridges, adjusting timing, intensity and descriptions for each part.
Also, Lyria 3 is available via the Gemini API, which makes it easy to integrate into applications and existing workflows.
Transparency and responsible use
Each track generated by Lyria 3 includes a digital watermark called SynthID. This helps identify and verify audio generated by Google AI, even after edits. It's a measure designed to maintain trust and traceability in AI-generated content.
Practical ideas and use cases
Content creator: generate 30-second intros for reels or podcasts with lyria-3-clip-preview.
Startups and marketing: produce jingles or background music for campaigns without relying on expensive libraries.
Game developers: generate dynamic soundtracks and adaptive sections for levels and scenes.
Composers and producers: use Composer mode to prototype structures and then polish them in the studio.
Want to try quickly? Use the Clip variant for prototypes and move to Pro when you need full-length tracks with more detail.
Final thoughts
Lyria 3 isn't just another automatic tool. It's designed so AI complements the creative process, offering fine control and results that can approach professional production. The question now is how you'll integrate it into your projects: a jingle, your game's soundtrack, or the demo for your next album?
