Media & Culture

Lyria 3 Google Deepmind's music generator

The new model creates 30-second tracks from text prompts and allows detailed instrument-level editing.

Deep Dive

Google DeepMind has launched Lyria 3, its latest AI music generation model. Key specs include a 128k token context window for longer musical coherence and the ability to generate 30-second tracks from text descriptions. Users can now create complex, multi-instrument compositions and make granular edits to individual musical elements like drums or melody, moving beyond simple prompt-to-audio generation.

Why It Matters

Enables musicians and creators to rapidly prototype and refine complex musical ideas with AI-assisted precision.