Google’s new AI turns text captions into music
MusicLM is reportedly capable of generating songs of "significant complexity"
Researchers at Google have been working on a new AI tool called MusicLM that can turn text captions into music.
An academic paper obtained by TechCrunch reveals that the tool was trained on a dataset of 280,000 hours of music and is capable of generating songs of "significant complexity", which range from “enchanting jazz song with a memorable saxophone solo and a solo singer” to “Berlin ’90s techno with a low bass and strong kick.”
Music LM will have the potential for going beyond song clips to generate songs from existing melodies, which could be sung, whistled or hummed. However, TechCrunch point out that among the failings of the tool are its nonsensical 'lyrics'. TechCrunch also reports that Google currently has no immediate plans to release MusicLM.
Check out the academic article, which includes examples of the tool in action. Captions veer from "melodic techno" to "a fusion of reggaeton and electronic dance music, with a spacey, otherworldly sound" and other eerie clips.
In another example, Google researchers inputted the following instructions to MusicLM: “The main soundtrack of an arcade game. It is fast-paced and upbeat, with a catchy electric guitar riff. The music is repetitive and easy to remember, but with unexpected sounds..."
Revisit our AI Futures series to understand how artificial intelligence will change music, learn about the impact of AI on DJing, and see how artificial intelligence is shaping music production.