Visualizing and Understanding Self-attention based Music Tagging
Sound
2019-11-12 v1 Audio and Speech Processing
Abstract
Recently, we proposed a self-attention based music tagging model. Different from most of the conventional deep architectures in music information retrieval, which use stacked 3x3 filters by treating music spectrograms as images, the proposed self-attention based model attempted to regard music as a temporal sequence of individual audio events. Not only the performance, but it could also facilitate better interpretability. In this paper, we mainly focus on visualizing and understanding the proposed self-attention based music tagging model.
Cite
@article{arxiv.1911.04385,
title = {Visualizing and Understanding Self-attention based Music Tagging},
author = {Minz Won and Sanghyuk Chun and Xavier Serra},
journal= {arXiv preprint arXiv:1911.04385},
year = {2019}
}
Comments
Machine Learning for Music Discovery Workshop (ML4MD) at ICML 2019