Recording and production technology over the past, what, 60 years has steadily bled the human from vocal performance. It started with bits of reverb and echo to mask bad or non-perfect singing, arced upward into vocorders and stylistic gimmicks, and, now, AutoTune is de rigueur in pop music for even the best voice. There are no good singers left because we won’t let them.I don’t believe this is a fad. It’s the natural progression of music that has become unmusic in its artificiality. Corporations killed pop music, sure, but it used formulas and automation and software as its weapons. Perhaps it won’t even need actual singers in the future and we can remove that last human from the equation. Researchers at the University of Tokyo have taken us closer than ever with a process that may prove to be at least the foundation of 100-percent computer-generated vocals that are aesthetically, um, practical?Programs like VOCALOID can come close, creating singing with the input of lyrics and a score, and it’s garnered a good bit of respect in Japan, but not so much in the West. It’s also still not entirely free from human voices, getting a large share of its attention for its ability to mimic the real voices of pop stars. It also requires intense participation from a human user.The problem seems to be in creating frequency curves that are sufficiently optimized enough to shed their synthetic markers. The UT takes an evolutionary approach, reaching an acceptable curve by essentially breeding curves into new frequency curves and using a sort of natural selection to find the best. It’s based on Interactive Evolutionary Computation, which is basically art-by-computer with human guidance rather than the above intense participation.Results are positive, so expect our computer composer/performer pop stars of the future to try and start selling you awful shit in the very near future.Reach this writer at michaelb@motherboard.tv.