Imagine composing a flawless, studio-quality music track in mere seconds simply by describing it in words or uploading an image or video.
Google’s latest breakthrough, Lyria 3, transforms this imagination into reality, redefining how creators produce, remix, and visualize music.
On Wednesday, Google unveiled the integration of its advanced Lyria 3 model into Gemini, allowing users to generate and remix 30-second music tracks from simple prompts. This upgrade expands Gemini’s existing multimodal capabilities and powers YouTube’s new “Dream Track” feature for Shorts.
Creators can craft music from straightforward textual prompts, such as a “comical R&B slow jam,” or exercise detailed control over tempo, drumming styles, and emotional tone.
Lyria 3 transcends conventional text inputs, enabling tracks to be generated from uploaded images or videos. Remarkably, these compositions can also be paired with AI-generated album artwork, produced by Google’s Nano Banana image model, offering a fully immersive creative experience.
According to a Google blog post, Lyria 3 represents a major leap in producing musically sophisticated tracks, featuring professional-grade arrangements and automated lyric generation.
While output is currently limited to 30-second clips, the model affords granular manipulation of individual song components.
All Lyria 3 creations incorporate Google’s SynthID watermarking, ensuring transparency and traceability through the SynthID Detector, introduced at Google I/O 2025.
While the instrumental elements are highly realistic, early feedback notes that AI-generated lyrics occasionally lack the natural fluidity of human composition.
The Lyria 3 model has now rolled out globally for Gemini users aged 18 and above, supporting prompts in English, Spanish, German, French, Hindi, Japanese, Korean, and Portuguese.
With Lyria 3, Google pushes the boundaries of AI-driven music creation, offering an unprecedented fusion of innovation, interactivity, and creative freedom.
