Google has developed an impressive AI system that can generate music of any genre given a text description. Called MusicLM, this system surpasses earlier algorithms in terms of audio quality and the ability to faithfully recreate descriptions. Samples have been created from a variety of genres including meditation music, electronic music, and even rap, and the system can be trained on a combination of pictures and captions as well as specific instruments and genres.
However, Google has no plans to release this system due to the ethical challenges it poses. For example, there is the risk of misappropriating creative content and the potential for copyright infringement of material used to train the AI. Furthermore, using AI-generated music in a commercial setting is currently a legal grey area.
Despite these risks, the research paper written by thirteen academics underlines the potential of MusicLM to revolutionize the music industry. The paper announces the release of a public dataset, which contains over 5,500 music-text combinations, and could be used to train and evaluate other music-based AIs.
The implications of this technology are huge. MusicLM could be used to assist artists in writing and producing new songs, or even create entirely new genres of music. It could be used to create soundtracks for movies, television shows, and video games. It could even be used to create music tailored to specific places, epochs, or requirements, like motivational music for workouts.
Google’s MusicLM has the potential to revolutionize the music industry, and while there are still many ethical and legal issues to be worked out, it is an exciting development that could pave the way for a new era of musical creativity.
Source: Techcrunch
English 
French