We recently published another paper on the compositional hierarchical model. In an attempt to further generalise the model, we adjusted and applied it to rhythmic input.
The model showed promising results; while it was trained unsupervisedly, it identified basic rhythmic patterns. Moreover, it adjusted to live music (live performance) where it unsupervisedly discriminated between different meters. Quite awesome!
The full paper is available here: https://www.mdpi.com/2076-3417/10/1/178?type=check_update&version=1
No Comments