Google is expanding the multimodal capabilities of its Gemini AI. Following the viral Nano Banana image generator, the company is releasing Lyria 3, a new AI model designed to generate music. Users can describe the desired music in words, or provide a picture or video, and Lyria 3 will create a high-fidelity track with custom lyrics.
Lyria 3, developed by Google's DeepMind division, is now rolling out to the desktop web version of Gemini and will soon be available in the mobile app globally. The tool allows for the creation of 30-second soundtracks for any moment, from funny jingles to lo-fi beats.
Accessing the feature is simple: users click the tool picker in the Gemini chat box and select the new music option. The model works similarly to AI image or video generation, interpreting prompts to produce music. For those needing inspiration, templates with existing tracks are available for modification.
Notably, using Lyria 3 in Gemini is free. This move places Google in the competitive AI music generation space, currently led by companies like Suno, which has faced copyright controversies.