Google has developed a new artificial intelligence (AI) system called “MusicLM” capable of generating high-fidelity music in various genres from a text description. However, the company is aware of the potential risks involved and has decided not to release it to the public, according to a report by TechCrunch.
Although MusicLM is not the first AI system to be developed for music generation, it stands out from others, such as Riffusion and Dance Diffusion, due to its ability to produce songs with complex compositions and high sound quality. Previous AI systems have struggled to generate pieces of this calibre due to technical limitations and a lack of training data.
To train MusicLM, the system was exposed to a dataset of 280,000 hours of music, enabling it to generate coherent songs that meet the description of “significant complexity.” This was demonstrated in an academic paper by Cornell University, where examples of text descriptions that MusicLM was capable of generating music from included phrases such as “enchanting jazz song with a memorable saxophone solo and a solo singer” and “Berlin ’90s techno with a low bass and strong kick.”
