Lyria

Lyria is Google DeepMind’s music generation model, built to turn text prompts into produced audio tracks. The version available through Vertex AI (Lyria 2) generates instrumental music from text descriptions, handling a wide range of genres and picking up on the texture of individual instruments, while the streaming variant, Lyria RealTime, produces continuous music over a WebSocket connection that responds to live parameter changes. DeepMind pairs every clip with a SynthID watermark so the audio can be identified as AI-generated after the fact.

The model sits alongside Google’s other generative media tools, Imagen for images and Veo for video, as part of a broader push to bring studio-grade generation into Google Cloud’s Vertex AI platform for developers and into consumer products like the Gemini app. Its target users include musicians and producers working through the Music AI Sandbox, a tool built with artists to assist with composition, as well as developers building music features into their own apps through the API.