Lyria 2
audioby Google · Updated February 15, 2026
Lyria 2 is Google DeepMind's flagship music generation model, powering AI music features in YouTube and other Google products. It generates high-fidelity music with complex arrangements, realistic instruments, and natural vocals across genres. Lyria 2 includes built-in SynthID watermarking for responsible AI and supports text-to-music, melody-to-music, and style transfer workflows.
Best For
Prompting Tips
- 1Describe the genre, mood, and energy level clearly
- 2Reference specific decades or sub-genres for more targeted results
- 3Mention specific instruments you want prominently featured
- 4Describe the arc of the song: intro, build, chorus, outro
- 5Use terms like "radio-quality" or "studio-produced" for polished output
Syntax & Constraints
Natural language prompts. Available through Google AI Studio and YouTube AI tools. Supports text-to-music and melody conditioning. SynthID watermarked.
Build Prompts for Lyria 2
Other Gemini Models
Google's most capable model with 1M+ token context.
Fast, efficient model for high-volume tasks.
Google's most powerful reasoning model.
Google's highest quality image generation model.
Google's most advanced image model with 2K native resolution.
Google's high-fidelity video generation model.
Google's latest video model with native audio generation.
Google's reasoning model that converts sketches and images into 3D-printable STL models.