English

Improving DNN-based Music Source Separation using Phase Features

Sound 2018-07-17 v3 Machine Learning Audio and Speech Processing

Abstract

Music source separation with deep neural networks typically relies only on amplitude features. In this paper we show that additional phase features can improve the separation performance. Using the theoretical relationship between STFT phase and amplitude, we conjecture that derivatives of the phase are a good feature representation opposed to the raw phase. We verify this conjecture experimentally and propose a new DNN architecture which combines amplitude and phase. This joint approach achieves a better signal-to distortion ratio on the DSD100 dataset for all instruments compared to a network that uses only amplitude features. Especially, the bass instrument benefits from the phase information.

Keywords

Cite

@article{arxiv.1807.02710,
  title  = {Improving DNN-based Music Source Separation using Phase Features},
  author = {Joachim Muth and Stefan Uhlich and Nathanael Perraudin and Thomas Kemp and Fabien Cardinaux and Yuki Mitsufuji},
  journal= {arXiv preprint arXiv:1807.02710},
  year   = {2018}
}

Comments

7 pages, 9 figures, Joint Workshop on Machine Learning for Music at ICML, IJCAI/ECAI and AAMAS, 2018

R2 v1 2026-06-23T02:53:44.106Z