Google has launched Lyria 3, an artificial intelligence model designed for music generation. This new AI allows users to create short musical pieces directly from natural language prompts or by uploading images and videos.
Lyria 3, accessible via Google's Gemini app and YouTube's Dream Track feature, generates 30-second tracks. Users can specify genre, tempo, and lyric language. The generated music will include an imperceptible watermark created by Google's SynthID technology, verifiable through the Gemini app.
Improvements over the previous Lyria 2 model include automatic lyric generation and enhanced output quality and complexity. The Gemini app also features an AI image generator, Nano Banana, for creating cover art, with potential for Lyria 3 to be integrated into developer APIs in the future.
The model is currently available to adult users on Gemini's mobile client, with desktop access planned. Premium subscription tiers will offer increased usage limits. This launch intensifies competition in the AI music generation market, challenging startups like Suno Inc.