At a time when we still debate over whether artificial intelligence (AI) will one day outsmart humans, the technology continues to make significant advancements in various sophisticated tasks, ranging from driving cars to helping us fall asleep.
Now, researchers have used AI to create a black metal album, named "Coditany of Timeness", which sounds like a real human band. The album was generated by two musical technologists who used a deep learning software tool that was fed a musical album, before letting it process the same and produce an imitation of its own.
To create Coditany of Timeness, the software broke down an album called "Diotima" by Krallice, a New York-based black metal band, into small segments of audio. The short chunks of sound were then run through a neural network, which was asked to guess how the next section of the track would sound. If the guess work was right, the network would then enhance the paths of the neural network leading to the correct answer.
"Early in its training, the kinds of sounds it produces are very noisy and grotesque and textural," CJ Carr, one of the creators of the algorithm, told the Outline, adding that the AI eventually became more sophisticated after as many as five million guesses over the course of three days.
"As it improves its training, you start hearing elements of the original music it was trained on come through more and more," Carr said.
Coditany of Timeness is part of a side project by Carr and Zack Zukowski, a music producer. The duo has previously created many experimental projects, blending music and machine learning under the name Dadabots. Some of these projects include a series of Soundcloud bots that can automatically download tracks, remix them and upload new versions.
"While we set out to achieve a realistic recreation of the original data, we were delighted by the aesthetic merit of its imperfections. Solo vocalists become a lush choir of ghostly voices, rock bands become crunchy cubist-jazz, and cross-breeds of multiple recordings become a surrealist chimera of sound. Pioneering artists can exploit this, just as they exploit vintage sound production," Carr and Zukowski wrote in a related research paper.
Listen to the set of five AI-created tracks here.
