Lyria 3

#13 in Multimodal Models

google · v3 · seit Februar 2026 · 28× · zuletzt 30. Juni 2026

26

Momentum

Lyria 3 is Google's most advanced music generation model, capable of synthesizing high-quality audio tracks from text prompts. The model supports vocal generation in multiple languages, understands musical structure (rhythm, arrangement, instrumentation), and produces 30-second tracks, while the Pro variant generates compositions up to 3 minutes long with structural awareness of intros, verses, choruses, and bridges.

Sources (28)