Google has introduced an innovative music-generation feature within the Gemini app, powered by DeepMind’s Lyria 3 model, which is currently in beta testing. This new capability allows users to create original songs simply by providing natural language prompts. The outcome is a 30-second track complete with lyrics and custom cover art, and users can even generate music inspired by uploaded photos or videos.
Lyria 3 is designed to offer more realistic musical compositions and gives users the ability to control various elements of the creation process, including style, vocals, and tempo. Additionally, Google is broadening access to its Dream Track tool for YouTube creators globally, making it available to users beyond the United States.
To ensure transparency, all audio produced through this feature will carry a SynthID watermark, distinguishing it as AI-generated content. Furthermore, Gemini will have the ability to detect AI-generated music. This exciting feature is set to launch worldwide for adult users in multiple major languages.
