Supervised Audio Source Separation Based on Nonnegative Matrix Factorization with Cosine Similarity Penalty

Yuta IWASE; Daichi KITAMURA

doi:10.1587/transfun.2021EAP1149

Open Access
Supervised Audio Source Separation Based on Nonnegative Matrix Factorization with Cosine Similarity Penalty

Yuta IWASE, Daichi KITAMURA

Full Text Views

97

Cite this

Free PDF (2.3MB)

Summary :

In this study, we aim to improve the performance of audio source separation for monaural mixture signals. For monaural audio source separation, semisupervised nonnegative matrix factorization (SNMF) can achieve higher separation performance by employing small supervised signals. In particular, penalized SNMF (PSNMF) with orthogonality penalty is an effective method. PSNMF forces two basis matrices for target and nontarget sources to be orthogonal to each other and improves the separation accuracy. However, the conventional orthogonality penalty is based on an inner product and does not affect the estimation of the basis matrix properly because of the scale indeterminacy between the basis and activation matrices in NMF. To cope with this problem, a new PSNMF with cosine similarity between the basis matrices is proposed. The experimental comparison shows the efficacy of the proposed cosine similarity penalty in supervised audio source separation.

Publication: IEICE TRANSACTIONS on Fundamentals Vol.E105-A No.6 pp.906-913

Publication Date: 2022/06/01

Publicized: 2021/12/08

Online ISSN: 1745-1337

DOI: 10.1587/transfun.2021EAP1149

Type of Manuscript: PAPER

Category: Engineering Acoustics

Authors

Yuta IWASE
Kagawa College
Daichi KITAMURA
Kagawa College

Keyword

audio source separation, nonnegative matrix factorization, orthogonality, cosine similarity

Cite this

Copy

Yuta IWASE, Daichi KITAMURA, "Supervised Audio Source Separation Based on Nonnegative Matrix Factorization with Cosine Similarity Penalty" in IEICE TRANSACTIONS on Fundamentals, vol. E105-A, no. 6, pp. 906-913, June 2022, doi: 10.1587/transfun.2021EAP1149.
Abstract: In this study, we aim to improve the performance of audio source separation for monaural mixture signals. For monaural audio source separation, semisupervised nonnegative matrix factorization (SNMF) can achieve higher separation performance by employing small supervised signals. In particular, penalized SNMF (PSNMF) with orthogonality penalty is an effective method. PSNMF forces two basis matrices for target and nontarget sources to be orthogonal to each other and improves the separation accuracy. However, the conventional orthogonality penalty is based on an inner product and does not affect the estimation of the basis matrix properly because of the scale indeterminacy between the basis and activation matrices in NMF. To cope with this problem, a new PSNMF with cosine similarity between the basis matrices is proposed. The experimental comparison shows the efficacy of the proposed cosine similarity penalty in supervised audio source separation.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.2021EAP1149/_p

Copy

@ARTICLE{e105-a_6_906,
author={Yuta IWASE, Daichi KITAMURA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Supervised Audio Source Separation Based on Nonnegative Matrix Factorization with Cosine Similarity Penalty},
year={2022},
volume={E105-A},
number={6},
pages={906-913},
abstract={In this study, we aim to improve the performance of audio source separation for monaural mixture signals. For monaural audio source separation, semisupervised nonnegative matrix factorization (SNMF) can achieve higher separation performance by employing small supervised signals. In particular, penalized SNMF (PSNMF) with orthogonality penalty is an effective method. PSNMF forces two basis matrices for target and nontarget sources to be orthogonal to each other and improves the separation accuracy. However, the conventional orthogonality penalty is based on an inner product and does not affect the estimation of the basis matrix properly because of the scale indeterminacy between the basis and activation matrices in NMF. To cope with this problem, a new PSNMF with cosine similarity between the basis matrices is proposed. The experimental comparison shows the efficacy of the proposed cosine similarity penalty in supervised audio source separation.},
keywords={},
doi={10.1587/transfun.2021EAP1149},
ISSN={1745-1337},
month={June},}

Copy

TY - JOUR
TI - Supervised Audio Source Separation Based on Nonnegative Matrix Factorization with Cosine Similarity Penalty
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 906
EP - 913
AU - Yuta IWASE
AU - Daichi KITAMURA
PY - 2022
DO - 10.1587/transfun.2021EAP1149
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E105-A
IS - 6
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - June 2022
AB - In this study, we aim to improve the performance of audio source separation for monaural mixture signals. For monaural audio source separation, semisupervised nonnegative matrix factorization (SNMF) can achieve higher separation performance by employing small supervised signals. In particular, penalized SNMF (PSNMF) with orthogonality penalty is an effective method. PSNMF forces two basis matrices for target and nontarget sources to be orthogonal to each other and improves the separation accuracy. However, the conventional orthogonality penalty is based on an inner product and does not affect the estimation of the basis matrix properly because of the scale indeterminacy between the basis and activation matrices in NMF. To cope with this problem, a new PSNMF with cosine similarity between the basis matrices is proposed. The experimental comparison shows the efficacy of the proposed cosine similarity penalty in supervised audio source separation.
ER -