Each year, participants from primarily European countries make a perfect surprise for the world in the Eurovision Song Contest. Fireworks, flashes, costume changes, key changes, wind machines, fake rain you gave the name, it probably happened in the last 4 years of the competition. But to fill this void, the AI song competition organized by Dutch public broadcaster VPRO. For the first time, 13 teams from Europe and Australia have acquired the creative power of artificial intelligence to generate the next Eurovision hit. And just like the real competition, you (the public) can help decide who gets the judge point and who gets away with the blue point.
Computer scientists have teamed up with folk musicians and machine learning experts to train AI systems to instantly compose 3-minute classics. Most teams started loading their systems with Eurovision music repositories and then helped
create new song tunes and melodies using various open-source tools, such as Google’s Magenta project. “In our case, it shows all the music from Eurovision,” Ashley Burgoyne, one of the members of the Dutch team, explained in a YouTube video. “We say to the computer: Listen, listen again and listen more, until it fully understands all the patterns and all the music. This way it will learn how to write new music on its own.
Not often, AI systems come up with some left-field concepts. The song generator Can AI Kick, created by the Dutch team, coined the completely new word “Abs” defining it as “atheist cloud”, although the German team Dadabots X Portrait X
also saw their system create fairly topical lyrics about infections and diseases. Such as “My species notices vital signs, pathways of infection, bodily fluids.” Naturally, the group combines these in a heavy-metallic love song between a
human girl and an AI boy during a deadly epidemic.
For the performance of these masterpieces, some groups chose to mix human voices with robotic words, while others were fully committed to the good, bad, and ugly of AI by publicly synthesized voices. The Australian chose to train their neural network to identify Australian bushfires, including audio samples of Cola, Kukaburra, and Tasmanian devils, including the Uncanny Valley, which had devastating consequences for its inhabitants.
Voters are asked to rate each AI song in four categories: Song, Lyrics, Originality, and Eurovision. The deadline to do so is May 10 and these scores will be combined with a panel of experts looking for elements such as data usage, song structure, and how much people are involved.
While the award may only be a slot on Dutch national radio, the goal of the whole project is to explore the creative possibilities and limitations of AI. Judging by the entries, next year’s Eurovision contestants may include some computer scientists.