DUBAI: Google has launched Lyria 3, a generative AI music model currently in final testing, that can be used via the Gemini website and app to create customized audio tracks.
Users can provide text prompts such as “an upbeat, modern Arabic fusion track for Ramadan” or “a massive, anthemic rock song with an emotive male singer.”
They can add images to their prompts and ask the model to generate a track that reflects the ideas within the pictures. They can also add lyrics or ask the model to generate them.
Lyria 3 then produces a 30-second track along with cover art generated by Google’s artificial intelligence image generator and editor, Nano Banana.
Google said the aim was not to create a musical masterpiece or for copying existing artists but to let users express themselves in unique ways. However, if a prompt specifies a particular artist, the model can draw inspiration their style while still creating an original track.
Lyria was launched in 2023 and is the company’s most advanced music generation model. SynthID, Google’s tool to watermark and identify AI-generated content, is embedded in all tracks it creates.
Users can also upload a file to check whether it was generated using Google AI. Gemini will examine it for SynthID and provide a response based on its analysis.
Lyria 3 is available in Arabic, English, Spanish, French, Hindi, Japanese, Korean and Portuguese, with more languages expected in the future. It will be available on the mobile app in the coming days.
