Moving from two releases created entirely with homemade instruments, including the acoustic EP Skin, experimental duo Emptyset took a more scientific approach with their 2019 full-length. Blossoms was conceived using a neural network-based artificial intelligence system. Emptyset's James Ginzburg and Paul Purgas spent 18 months developing a machine learning system with a network of programmers, and they supplied the system with several hours of music improvised on their own unique instruments, as well as some of the duo's previous recordings. The result inevitably sounds like an Emptyset record, but one that's been twisted further inside-out, with lurching quasi-rhythms that seem to swallow themselves and unruly amounts of eardrum-shredding distortion. The most engaging tracks are the ones that come closest to approximating an extremely damaged, sideways form of techno, such as the trippy multi-time grind of "Blossom" or the heavy, stuck-in-mud thump of "Axil." "Stem" appears to have some voices trapped in the matrix, providing the album's most haunting moment. "Blade" takes a break from all the cyborg pummeling, instead turning out to be a calm wash of droning fuzz that is actually quite pretty. "Clone" is somewhat similar, but its mesmerizing waves are much harsher, and "Bulb" is a slow murmur resembling an amplified scanner. Logically, Blossoms doesn't have the sort of strangely human touch of Emptyset's 2017 releases, but it's still a compelling, somewhat frightening hybrid of organic and synthetic processes.