Target Source Separation Based on Discriminative Nonnegative Matrix Factorization Incorporating Cross-Reconstruction Error

Kisoo KWON; Jong Won SHIN; Nam Soo KIM

doi:10.1587/transinf.2015EDL8114

Target Source Separation Based on Discriminative Nonnegative Matrix Factorization Incorporating Cross-Reconstruction Error

Kisoo KWON, Jong Won SHIN, Nam Soo KIM

Full Text Views

0

Share
Cite this

Summary :

Nonnegative matrix factorization (NMF) is an unsupervised technique to represent nonnegative data as linear combinations of nonnegative bases, which has shown impressive performance for source separation. However, its source separation performance degrades when one signal can also be described well with the bases for the interfering source signals. In this paper, we propose a discriminative NMF (DNMF) algorithm which exploits the reconstruction error for the interfering signals as well as the target signal based on target bases. The objective function for training the bases is constructed so as to yield high reconstruction error for the interfering source signals while guaranteeing low reconstruction error for the target source signals. Experiments show that the proposed method outperformed the standard NMF and another DNMF method in terms of both the perceptual evaluation of speech quality score and signal-to-distortion ratio in various noisy environments.

Publication: IEICE TRANSACTIONS on Information Vol.E98-D No.11 pp.2017-2020

Publication Date: 2015/11/01

Publicized: 2015/08/19

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2015EDL8114

Type of Manuscript: LETTER

Category: Speech and Hearing

Authors

Kisoo KWON
  Seoul National University
Jong Won SHIN
  Gwangju Institute of Science and Technology
Nam Soo KIM
  Seoul National University

Keyword

nonnegative matrix factorization, discriminative basis, cross-reconstruction error

Cite this

Copy

Kisoo KWON, Jong Won SHIN, Nam Soo KIM, "Target Source Separation Based on Discriminative Nonnegative Matrix Factorization Incorporating Cross-Reconstruction Error" in IEICE TRANSACTIONS on Information, vol. E98-D, no. 11, pp. 2017-2020, November 2015, doi: 10.1587/transinf.2015EDL8114.
Abstract: Nonnegative matrix factorization (NMF) is an unsupervised technique to represent nonnegative data as linear combinations of nonnegative bases, which has shown impressive performance for source separation. However, its source separation performance degrades when one signal can also be described well with the bases for the interfering source signals. In this paper, we propose a discriminative NMF (DNMF) algorithm which exploits the reconstruction error for the interfering signals as well as the target signal based on target bases. The objective function for training the bases is constructed so as to yield high reconstruction error for the interfering source signals while guaranteeing low reconstruction error for the target source signals. Experiments show that the proposed method outperformed the standard NMF and another DNMF method in terms of both the perceptual evaluation of speech quality score and signal-to-distortion ratio in various noisy environments.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2015EDL8114/_p

Copy

@ARTICLE{e98-d_11_2017,
author={Kisoo KWON, Jong Won SHIN, Nam Soo KIM, },
journal={IEICE TRANSACTIONS on Information},
title={Target Source Separation Based on Discriminative Nonnegative Matrix Factorization Incorporating Cross-Reconstruction Error},
year={2015},
volume={E98-D},
number={11},
pages={2017-2020},
abstract={Nonnegative matrix factorization (NMF) is an unsupervised technique to represent nonnegative data as linear combinations of nonnegative bases, which has shown impressive performance for source separation. However, its source separation performance degrades when one signal can also be described well with the bases for the interfering source signals. In this paper, we propose a discriminative NMF (DNMF) algorithm which exploits the reconstruction error for the interfering signals as well as the target signal based on target bases. The objective function for training the bases is constructed so as to yield high reconstruction error for the interfering source signals while guaranteeing low reconstruction error for the target source signals. Experiments show that the proposed method outperformed the standard NMF and another DNMF method in terms of both the perceptual evaluation of speech quality score and signal-to-distortion ratio in various noisy environments.},
keywords={},
doi={10.1587/transinf.2015EDL8114},
ISSN={1745-1361},
month={November},}

Copy

TY - JOUR
TI - Target Source Separation Based on Discriminative Nonnegative Matrix Factorization Incorporating Cross-Reconstruction Error
T2 - IEICE TRANSACTIONS on Information
SP - 2017
EP - 2020
AU - Kisoo KWON
AU - Jong Won SHIN
AU - Nam Soo KIM
PY - 2015
DO - 10.1587/transinf.2015EDL8114
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E98-D
IS - 11
JA - IEICE TRANSACTIONS on Information
Y1 - November 2015
AB - Nonnegative matrix factorization (NMF) is an unsupervised technique to represent nonnegative data as linear combinations of nonnegative bases, which has shown impressive performance for source separation. However, its source separation performance degrades when one signal can also be described well with the bases for the interfering source signals. In this paper, we propose a discriminative NMF (DNMF) algorithm which exploits the reconstruction error for the interfering signals as well as the target signal based on target bases. The objective function for training the bases is constructed so as to yield high reconstruction error for the interfering source signals while guaranteeing low reconstruction error for the target source signals. Experiments show that the proposed method outperformed the standard NMF and another DNMF method in terms of both the perceptual evaluation of speech quality score and signal-to-distortion ratio in various noisy environments.
ER -