These Tunes Were Written By Artificial Intelligence. Can You Hear the Difference?
An A.I. learned to compose traditional folk tunes that sound anything but digital.
PHOTO CREDIT: Getty Images
There's nothing very digital about the sound of traditional folk music from Ireland and Britain, but for the past few months an album of folk tracks written by an artificial intelligence has been passing as a regular, human-composed record in the wild.
The collection of songs "Let's Have Another Gan Ainm" on Soundcloud is credited to the artist "O Conaill Family and Friends," but it's actually the project of Bob Sturm and other researchers at KTH Royal Institute of Technology.
Sturm trained a recurrent neural network using on online database of folk music and the system then "learned" how to compose its own tunes.
"The resulting computer models show some ability to repeat and vary patterns in ways that are characteristic of this kind of music," Sturm says. "It was not programmed to do this using rules - it learned to do so because these patterns exist in the data we fed it."
The AI-written compositions were then turned over to a group of professional Irish traditional musicians to record an album, which was subsequently released under the aforementioned fictional name in order to solicit unbiased reviews and opinions from both professional critics and the public.
"We had to make up a story about the album's origins in order to avoid the bias that can result when someone believes a creative product was created by a computer," Sturm says. "And now that we have reviews, we are revealing the true origins of the album."
One of the tracks can be heard above. See if it sounds composed by a lifeless hunk of computer code to you or not.
"People have not suspected a thing about the tunes, that they were computer-generated," Sturm says.
He adds that the idea behind the work is not to replace human composers with computers, but to create new tools.
"Our work with many collaborators, such as composer Oded Ben-Tal at Kingston University in the UK, and professional musicians, has also shown how the models can serve a wider purpose: as useful partners in creating new music," Sturm says.