In a move that pushes the boundaries of creative technology, Google has officially integrated music generation into the Gemini app. Powered by Lyria 3, Google DeepMind’s most advanced audio model to date, the feature allows users to transform simple text prompts or even personal photos into high-quality, 30-second soundtracks. Joël Yawili, Senior Product Manager for the Gemini app, noted that since the app’s launch, the focus has been on encouraging creative expression through images and video, but this latest step into custom music generation marks a significant evolution. The update shifts how users interact with AI, moving from text-based assistance to a full-fledged creative companion capable of composing rhythm, melody, and lyrics in seconds.
Lyria 3 is designed for personal expression rather than just background noise, introducing major upgrades over its predecessors. One of the most significant changes is that users no longer need to provide their own lyrics, as Gemini now crafts verses automatically based on the provided story or vibe. Additionally, the model provides much deeper creative control, allowing users to specify exact styles, vocal tones, and tempos to fine-tune the output. These tracks are more realistic and structurally complex than ever before, capable of handling everything from a comical R&B slow jam about a lost sock to a nostalgic Afrobeat track celebrating a mother’s home-cooked plantains.
One of the most captivating features of the rollout is the Image-to- Track capability. Users can upload a photo or video, such as a snapshot of a pet on a hike or a family gathering, and Gemini will analyse the visual context to compose a fitting soundtrack. This makes the tool perfect for creating unique social media content, personalised digital cards, or simply reliving memories through a new medium. While the music is being composed, the app also generates custom cover art via Nano Banana, making the final product ready for immediate sharing
With great creative power comes the need for transparency, and Google has addressed this by embedding every generated track with SynthID, an imperceptible digital watermark developed by Google DeepMind. To combat the rise of AI-generated misinformation, Google has also expanded Gemini’s verification tools, allowing users to upload an audio file and ask the AI if it was generated by Google. Gemini then scans for the SynthID watermark and uses its own reasoning to provide a response.
The reach of Lyria 3 extends beyond the standalone app into YouTube Dream Track, enhancing the quality of soundtracks for Shorts creators worldwide. By providing lyrical verses and vibey backing tracks on demand, Google is equipping a new generation of creators with professional-grade tools to elevate their short-form content. As Lyria 3 rolls out to users aged 18 and older across various languages, the message is clear: the future of AI is no longer just about what it can tell you, but what it can sing to you.
