The project, called ‘Relentless Doppelganger’, which was officially launched on Mar 24, 2019 via a YouTube channel uses an ‘Unconditional End-to-End Neural Audio Generation Model’ to generate samples that have been trained specifically around the genre of death metal.
The researchers used Canadian death metal band, Archspire, to help train the Neural Network into “learning” how to create death metal music.
In their quest to wipe out humans, the neural network actually generates some incredibly interesting music.
The team originally created SampleRNN to generate random rock music but have decided to try their hand at Death Metal.
“We use a modified SampleRNN architecture to generate music in modern genres such as black metal and math rock. Unlike MIDI and symbolic models, SampleRNN generates raw audio in the time domain. This requirement becomes increasingly important in modern music styles where timbre and space are used compositionally. Long developmental compositions with rapid transitions between sections are possible by increasing the depth of the network beyond the number used for speech datasets. We are delighted by the unique characteristic artifacts of neural synthesis.”
Speaking to Futurism.com, Carr said: "It seems the faster the blast beats, the more stable the music, The remarkable part is the high quality-to-shit ratio, Here, we livestream 100 percent of it"
Live stream Neural Network Death Metal here.