Google announced Wednesday that it is adding music generation functionality to its Gemini app. The company is using DeepMind’s Lyria 3 music generation model to power this feature, which is still in beta.
To use this feature, describe the song you want to create and the app will generate a track with lyrics. For example, if you ask Gemini to create a “comic R&B slow jam about finding the right socks,” the app will generate a 30-second track with cover art created by Nano Banana.
Google says you can also upload photos and videos, and its AI-powered tools will create songs that match the mood of your media files.
The company says Lyria 3 improves on the previous generation model, allowing you to create more realistic and complex music tracks. Users can also change and control other elements such as style, vocals, and tempo.
Google is rolling out Lyria 3 to the Gemini app and making the model available to YouTube creators through YouTube’s Dream Tracks feature. The Dream Tracks feature is a tool that helps creators create AI-generated tracks. This option was previously only available to YouTube creators in the United States. However, with this release, Google is expanding the availability of Dream Track globally.
Google said it can’t imitate an artist perfectly, but if you add an artist’s name to the prompt, Gemini will create tracks with a similar style and feel. (It is unclear whether generation makes it easier for others to decipher a particular artist’s musical style.)
“Music generation with Lyria 3 is designed for original expression, not for imitating existing artists. When a prompt specifies the name of a particular artist, Gemini takes this as broad creative inspiration and creates tracks that share a similar style and feel. It also provides filters to check the output against existing content,” the company said in a blog post.
Google noted that all songs created with Lyria 3 models will include a SynthID watermark to identify AI-generated content. The company said it will also add the ability to identify AI-generated music using SynthID within Gemini. Users can upload a track and ask Gemini if it was generated by AI.
Music generation supports English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese and is rolled out to all Gemini users 18 years and older worldwide.
AI-generated music has evoked mixed emotions among artists and listeners. Meanwhile, companies like YouTube and Spotify are deploying AI and signing deals with music labels to monetize the music they generate. Meanwhile, AI model and tool companies are facing lawsuits from the music industry over copyrights to their training materials. Platforms like Deezer are exposing tools to mark AI-generated music to curb unauthorized streams of this type of music.
