Gemini Can Now Create Music for You
▼ Summary
– Google has launched Lyria 3, an AI music generation tool integrated into Gemini that creates 30-second songs from text, photo, or video prompts.
– The new version adds automatic lyric generation and offers users more control over musical style, vocals, and other track details.
– Google emphasizes the tool is for original expression, not mimicking artists, and includes filters to avoid copyright infringement, though they may not be perfect.
– Lyria 3 is an update to a model first launched in 2023 and previously used in YouTube experiments like Dream Track for creating Shorts soundtracks.
– The tool is rolling out globally starting February 18 for adult Gemini users on desktop, with mobile access to follow, and supports multiple major languages.
Ever wanted to compose a tune but felt you lacked the skill or creative spark? Google’s latest AI update, Lyria 3, now lets anyone generate a 30-second song directly within Gemini using a simple text prompt, photo, or video. This new iteration of the music creation model is far more accessible than its predecessors, joining other creative tools like Nano Banana for images and Veo for video in the Gemini Tools menu. The process requires little more than a short description to produce a complete track with AI-generated lyrics.
Google showcased the feature with whimsical examples, including an R&B song about socks finding their match in a washing machine and an afrobeat track reminiscing about cooking plantains with one’s mother. The company emphasized that users don’t need to supply their own words; the system crafts lyrics automatically based on the prompt. Senior product managers noted the goal is to provide a fun and unique avenue for personal expression. Each generated song even comes with complimentary cover art created by the Nano Banana image AI.
Lyria first debuted in 2023 and was later integrated into YouTube experiments like Dream Track, which allowed creators to produce 30-second soundtracks for Shorts in the style of certain artists, and the Music AI Sandbox, a toolkit for musicians. The new Lyria 3 model introduces several upgrades, most notably the automatic lyric generation and greater user control over musical style, vocals, and other specifics. Google claims it can also produce more realistic and complex musical arrangements. While Dream Track is being updated with Lyria 3, songs are currently capped at 30 seconds in length.
A significant focus for Google with this release is navigating the complex landscape of music copyright. The original Lyria model, used in Dream Track, was trained on licensed material from artists including T-Pain, Demi Lovato, and Sia. For Lyria 3, Google states the system is programmed to be “very mindful of copyright” and the agreements made with partnering musicians. The company asserts the tool is designed for original expression rather than mimicking existing artists. If a user names a specific musician in their prompt, Gemini will treat it as broad creative inspiration to generate a track in a similar style or mood, not an imitation.
Google has also implemented filters intended to prevent the AI from generating content too similar to existing songs, though it acknowledges these safeguards may not be perfect. The company asks users to report any output that might infringe on copyright. This cautious approach highlights the ongoing industry-wide effort to balance AI innovation with intellectual property rights.
For those interested in experimenting with this new form of AI-assisted music creation, Lyria 3 began its global rollout on February 18. It is available to all Gemini users aged 18 and over in several languages, including English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese, with more to be added. The feature is launching on desktop first, with access via the Gemini mobile app following in the subsequent days.
(Source: The Register)
