Developing with the Latest Music Generation Model Lyria 3

Google has unveiled Lyria 3 and Lyria 3 Pro, its latest music generation models capable of high-quality music and vocal generation. Developers can use these models through the Gemini API and Google AI Studio.

The models come in two variants depending on the production purpose.

• Lyria 3 Pro: A premium model that generates tracks around 3 minutes long, with studio-quality structural completeness.
• Lyria 3 Clip: A model optimized for speed and bulk requests, generating high-quality 30-second clips suitable for prototyping or social media assets.

Lyria 3 supports sophisticated control through natural language prompts. It enables Tempo conditioning, Time-aligned lyrics that control the flow of the lyrics, and multimodal Image-to-music input that determines the mood of the music through an image.

All generated tracks include a SynthID digital watermark, designed to transparently identify and verify AI-generated audio.