Google Integrates Lyria 3.5 Music AI Model into Gemini and Developer APIs
Google has officially deployed its Lyria 3.5 music generation model to the Gemini app and Gemini API, moving the technology from experimental labs to mainstream access. The update enables users and developers to generate full-length, multi-minute songs complete with verses, bridges, and choruses directly from text and visual prompts. This launch significantly strengthens Google's generative AI footprint in audio production by giving developers direct API access via Google AI Studio and Vertex AI. By supporting image inputs and embedding robust audio watermarking, Google aims to streamline creative workflows while addressing growing legal and origin-tracking concerns in AI music. The release features two models: Lyria 3.5 for full multi-minute songs (supporting MP3 and WAV formats) and Lyria 3 Clip for 30-second loops. Developers can input up to 10 images alongside text prompts to guide music generation, and all outputs automatically include invisible SynthID watermarks that resist compression and resampling.
## BACKGROUND
Lyria is a family of music generation models developed by Google DeepMind designed to produce high-quality vocals, instrumentation, and full tracks. To safeguard AI-generated content, Google developed SynthID, a technology that embeds imperceptible digital watermarks directly into audio waveforms to verify provenance without degrading sound quality.