Google has added a new AI-powered music-generation feature to its Gemini app using DeepMind’s latest model, Lyria 3.
With this update, users can now ask Gemini to compose original 30-second music tracks just by typing a prompt describing what they want or by uploading an image or video to set the mood for the song.
The tool works by interpreting text descriptions like genre, mood, or theme and then generating music complete with instrumental sound and lyrics. It also produces custom cover art for each track using Google’s Nano Banana image model and embeds an AI watermark (SynthID) so that people can identify it as AI-generated content.
In a Google Workspace update blog titled ‘Create custom soundtracks with Lyria 3 in the Gemini app’, it noted that end users who have access to the Gemini app will receive access to Lyria 3 automatically. To get started, select ‘Create music’ from the tools menu.
This music generation feature is currently in beta and available in the Gemini app for desktop and mobile in multiple languages (English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese) for users 18 and over. Subscribers to Google AI Plus, Pro, and Ultra plans get higher usage limits.
In addition to the Gemini app itself, Google is integrating Lyria 3 into YouTube’s Dream Track tool, letting creators generate custom soundtracks for Shorts and other video content.
Google noted that this feature is meant for fun and personal creative expression, not professional music production, and it includes safeguards so it doesn’t try to copy specific artists’ work.
