Skoltech Center for Energy Systems is pleased to invite you to the lecture «Structure Preserving Dimension Reduction for Musical Signals», which is a part of the seminar “Modern machine learning and application”.

Speaker

Ekaterina Krymova, Research assistant at University Duisburg-Essen, Germany.

Abstract

In the talk we will discuss mathematical problems which arise in cochlear implant music transmission. Also a novel unsupervised segmentation procedure for music signals will be presented which relies on an explained variance criterion in the eigenspace of the constant-Q spectral domain. It is compared to a segmentation based on equidistant boundaries. The results demonstrate that the proposed segmentation procedure gives an improvement in terms of signal-to-artefacts ratio in comparison to corresponding equidistant boundaries segmentation.