Google has launched Lyria 3, its advanced generative music model, in the Gemini app, enabling users to create 30-second tracks from text prompts or images.
Lyria 3 Model
Senior Product Managers Joël Yawili and Myriam Hamed Torres announced the feature, which generates lyrics, styles, vocals, and tempos based on descriptions like “a fun afrobeat track about childhood memories of home-cooked plantains.”
Read Also: Nigerian’s $150m Suit Against Google, GoDaddy Adjourned to April 22
Key upgrades include auto-generated lyrics, user control over elements, and more realistic, complex audio.
Usage Options
Text-to-track: Prompt genres, moods, or memories, e.g., “nostalgic afrobeat for my mother’s plantain meals with African vibe.”
Image/video-to-track: Upload photos/videos for mood-matched songs, e.g., “track about my dog on a hike.”
Tracks include custom Nano Banana cover art for easy sharing. Available for users 18+ in English, German, Spanish, French, Hindi, Japanese, Korean, Portuguese; Google AI subscribers get higher limits.
Verification and Responsibility
All outputs embed SynthID watermark; users can verify uploaded audio for Google AI origin.
Read Also: Africa’s AI Future: Google Opens Applications for Milestone 10th Startup Accelerator
Developed with music community input since 2023, Lyria 3 avoids mimicking artists—prompts inspire style only—with filters, reporting, and policy enforcement against IP violations.
Also enhances YouTube Shorts soundtracks via Dream Track for creators.
Access at gemini.google.com.
