Gemini can now generate a 30-second approximation of what real music sounds like (www.engadget.com)

Google has unveiled a significant upgrade to its Gemini model with the introduction of Lyria 3, enabling users to generate 30-second music tracks based on simple prompts or remix existing songs. This advancement enhances Gemini's existing capabilities in text, image, and video generation and is set to be integrated into YouTube's "Dream Track" feature. Users can create diverse music styles, from a comical R&B jam to complex backing tracks, and can even customize aspects like tempo and drumming style. Remarkably, Lyria 3 also allows for music generation based on images or videos, accompanied by album art created with Google's Nano Banana model. Lyria 3 represents a leap forward in creating realistic audio content, granting creators more control over track elements and automatically generating song lyrics. While current outputs are limited to 30 seconds, the potential for longer clips and broader applications, such as integration with Google Messages, is promising. Additionally, all AI-generated tracks will be watermarked with Google's SynthID for easy identification, ensuring that users are aware of their AI origins. Though the instrumental compositions are commendable, some lyrical outputs may seem quirky or awkward, reflecting the ongoing evolution of AI in music generation.