Generating Black Metal and Math Rock: Beyond Bach, Beethoven, and Beatles
This work addresses music generation for niche genres where timbre and spatial effects are compositionally important, though it is incremental as it builds on existing SampleRNN methods.
The paper tackled generating music in modern genres like black metal and math rock using a modified SampleRNN architecture, resulting in the ability to produce raw audio with unique neural synthesis artifacts and handle long compositions with rapid transitions by increasing network depth.
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.