(Credit: Google)
After image and video generation, it’s time for music generation on Google’s Gemini chatbot. The company just announced its latest music-generation model, Lyria 3, which will enable Gemini users to generate 30-second audio clips from textual and visual prompts.
Getting started is as easy as generating images using Nano Banana. Click the “Create Music” button below the prompt box. You can choose from a set of tunes based on genres, such as 90s rap, Latin pop, R&B romance, Afropop, and more, or describe the mood, style, vocals, and tempo you’d like your track to have.
(Credit: Google)Lyria 3 will also add lyrics based on your description, which can contain images for reference. Google’s example for an image-based prompt says: “Use these photos to create a track about my dog Duncan on a hike in the woods.” Once you hit Enter, Gemini will generate a 30-second clip with album art from Nano Banana. You’ll be able to download the audio track and also share it using a link.
(Credit: Google)The goal here isn’t “to create a musical masterpiece, but rather to give you a fun, unique way to express yourself,” Google says. The model is “designed for original expression, not for mimicking existing artists," and its filters should detect existing content.
Additionally, all songs generated by Gemini will carry Google’s Synth ID watermark. If you are unsure whether a clip was generated by Gemini, you can upload it to Gemini and ask the chatbot to verify.
AI song generation isn’t entirely new. Platforms like Suno and Udio have been at it for a couple of years, but they ran into copyright issues with major music labels like Sony, Universal Music, and Warner Bros. Suno settled its case with Warner, and Udio settled its case with Warner and Universal by signing licensing deals.
YouTube was reportedly also considering licensing its AI tools in 2024. Earlier this month, it introduced AI Playlist generation for YouTube Premium and YouTube Music Premium.
Lyria 3 in Gemini will roll out to all users aged 18+ on desktop starting today, and on the mobile app in the coming days. Gemini’s music generation currently produces tracks in English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese. Google plans to add support for additional languages in the future.
Google didn’t specify the limit for free users, but said that AI Plus, Pro, and Ultra subscribers will enjoy higher limits. The model will also be made available in YouTube's Dream Track tool for creating Shorts.