A new AI tool called MusicLM has been released by American tech giant Google that allows users to create musical notation from text inputs.
The program, which a group of Google researchers developed, is comparable to ChatGPT, which transforms text commands into stories, or DALL-E, which creates visuals from written suggestions.
The AI algorithm can be used to “transform text input into seconds, and even minutes, of music, as well as turn, hummed melodies into various instruments,” the company claims.
“A series of written instructions is provided, and this produces the audio. According to the researchers, these affect how the model continues the semantic tokens derived from the previous caption.”
A “story mode” example is available in addition to the music modulation capability. In this sample, the model is essentially given various text inputs with time durations for each form of music that needs to be made.