Google is rolling out a new Gemini feature enabling people to generate music with a text prompt.
Dubbed Lyria 3, the new generative AI model from Google’s DeepMind is rolling out globally in the Gemini app starting Feb. 18. You can also access it via the Gemini website.
Google claims Lyria 3 can generate up to 30 seconds of music from a simple text description, or even from a photo or video. The company also says people can upload their own tracks to help shape what Gemini generates.
Lyria brings three main improvements over Google’s previous audio generation models. First, Lyria 3 can generate lyrics based on user prompts. The newest model also gives people more control over various elements, like style, vocals, and tempo. Finally, Lyria 3 can generate “more realistic and musically complex tracks.”
Along with generating 30-second tracks, Gemini leverages Google’s Nano Banana model to generate cover art. In a blog post, Google framed this as a way for people to express themselves, rather than a way to create musical masterpieces.
However, Google’s intent for Lyria 3 might not mean much in the long run. We’ve already seen fake, AI-generated music proliferate on platforms like Spotify — more robust music generation models will likely only make that problem worse.
To that end, it’s worth noting that Google says all music generated via the Gemini app is embedded with the company’s SynthID watermark, which helps identify AI-generated content. Google says anyone can upload a file to Gemini and ask if it was generated using Google AI.
Google also says it designed Lyria for “original expression, not for mimicking existing artists” and says it has filters in place to check Lyria’s outputs against existing content. But in classic AI fashion, Google admits the approach “might not be foolproof” and asks people to report content that might violate rights.
Lyria 3 is available for all users over the age of 18 in English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese. Google plans to expand to more languages over time. The model is rolling out on desktop today, and the mobile app over the next several days. Finally, Google says subscribers to its AI plans will enjoy “higher limits” on music generation.
