Google's Gemini has expanded its capabilities, now allowing users to create music alongside its existing features such as text writing, image generation, and video editing. This latest addition utilizes the innovative Lyria 3 audio generation model, which has been integrated into Gemini and is now available for all users. Previously, Lyria was limited to select Google products, but this update marks its first widespread release.
Creating music with Lyria is as straightforward as generating images or videos. Users simply describe the type of music they want, and the AI handles the rest. For instance, you could request "a humorous R&B slow jam about a sock finding its pair" or "a sea shanty discussing the risks of AI slop"--the choice is yours.
Upon clicking the "Create music" button within the Gemini app, users can also remix existing tracks if they need inspiration. The presets cover a wide range of genres, from folk ballads to Latin pop, showcasing the diverse musical styles available. Additionally, users can upload an image or video to inspire the composition, allowing Lyria to craft music and lyrics that reflect the mood of the provided content. For example, you could upload pictures of your dog and receive a personalized tune about their adventures.
Currently, the generated tracks are limited to 30 seconds. While all Gemini users can create music, those with Plus, Pro, or Ultra subscriptions will enjoy higher usage limits, although specific details on these limits have yet to be clarified. According to Google, the goal of this feature is not to produce musical masterpieces but to offer a unique and enjoyable way for users to express themselves. While you won't be forming your own AI-generated band for streaming services, you can create entertaining tracks for personal enjoyment.
While I have not tested the feature personally, samples provided by Google appear somewhat generic, resembling the average sound derived from extensive audio training data--essentially, song genres distilled into their most common elements.
Google assures that all generated tracks will include invisible watermarks via SynthID, marking them as AI creations. Users can also upload audio tracks to Gemini for a SynthID verification. The updated Lyria 3 model is also being incorporated into the Dream Track music maker for YouTube Shorts creators. Currently, Lyria 3 is rolling out within Gemini, accessible to users aged 18 and over in multiple languages including English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese. Initially available on the web app, it will soon be integrated into the mobile app as well, with plans for future enhancements in quality and coverage.