Google has introduced AI-generated music to its Gemini app, expanding the platform’s creative capabilities beyond text, images, and video.
The new feature is powered by Lyria 3, a generative music model developed by Google DeepMind, and is currently rolling out in beta.
It allows users to generate short music tracks using text prompts or visual inputs such as photos and videos. According to Google, the feature is intended for casual creative use rather than professional music production.
How the Feature Works
Tracks generated in the Gemini app are limited to 30 seconds and may be instrumental or include AI-generated lyrics. Prompts can specify a genre, mood, or idea, or reference an uploaded image or video that Gemini uses as creative context.
Unlike earlier AI music tools, Lyria 3 does not require users to provide lyrics. The system generates them automatically and allows some control over musical elements such as tempo, vocal style, and overall tone.
Each generated track includes AI-created cover art and can be downloaded or shared via a link.
Earlier Lyria models were primarily available through developer platforms such as Google Cloud’s Vertex AI, rather than directly to Gemini users. Google says Lyria 3 improves on previous versions in several areas:
- Expanded control over musical structure and style
- Automatically generated lyrics based on prompts
- Increased realism and complexity in compositions
These changes aim to make the output sound more cohesive and less synthetic compared to earlier models, though the tracks remain short and relatively simple.
YouTube Shorts Integration
Lyria 3 is also being introduced to YouTube’s Dream Track feature, which allows creators to generate custom soundtracks for Shorts. The update supports both instrumental backing tracks and short lyrical segments. The feature is available in the U.S. and is gradually rolling out to creators in other regions.
All music generated in the Gemini app is embedded with SynthID, Google’s digital watermarking technology for identifying AI-generated content. Gemini’s existing verification tools have been expanded to include audio, enabling users to upload a file and check whether it contains Google’s AI watermark.
Google states that Lyria 3 is designed to generate original music rather than replicate existing songs or artists. If a prompt references a specific artist, Gemini interprets this as general stylistic inspiration instead of direct imitation. The company says it uses filters to reduce similarities to copyrighted works and allows rights holders to report potential issues.
The feature is available to Gemini users aged 18 and older in several languages in addition to English, including Spanish, French, German, Hindi, Japanese, Korean, and Portuguese. The rollout begins on desktop, with mobile availability following, and Gemini subscribers receive higher usage limits.
The addition of music generation comes as AI-powered creative tools are becoming more widely used and increasingly multimodal, especially for short-form and experimental content
As with other generative AI tools, how effectively these safeguards work in practice remains an open question, particularly as AI-generated music becomes easier to produce and distribute.
Leave a Reply