DIVERGENCE OPTIMIZATION IN NONNEGATIVE MATRIX FACTORIZATION WITH SPECTROGRAM RESTORATION FOR MULTICHANNEL SIGNAL SEPARATION

被引：0
作者：
Kitamura, Daichi [1 ]
Saruwatari, Hiroshi [1 ]
Nakamura, Satoshi [1 ]
Takahashi, Yu [2 ]
Kondo, Kazunobu [2 ]
Kameoka, Hirokazu [3 ]
机构：
[1] Nara Inst Sci & Technol, Nara 6300101, Japan
[2] Yamaha Corp, Shizuoka, Japan
[3] Univ Tokyo, Tokyo, Japan
来源：
2014 4TH JOINT WORKSHOP ON HANDS-FREE SPEECH COMMUNICATION AND MICROPHONE ARRAYS (HSCMA) | 2014年
关键词：
NMF; multichannel signal separation; music signal processing; spectrogram restoration; optimal divergence; AUDIO SOURCE SEPARATION; MIXTURES;
D O I：
暂无
中图分类号：
O42 [声学];
学科分类号：
070206 ; 082403 ;
摘要：
In this paper, we address an optimization issue for the divergence in supervised nonnegative matrix factorization with spectrogram restoration, which has been proposed for addressing multichannel signal separation. This method separates non-target components and reconstructs some missing data caused by preceding spatial clustering via supervised basis extrapolation. In our previous study, we only used a limited type of divergence, whereas the divergence selection is essential. Therefore, we extend this method to a more generalized form and give a theoretical analysis of the divergence optimization, where we reveal the trade-off between separation and extrapolation abilities.
引用
收藏
页码：92 / 96
页数：5
相关论文
共 43 条
[21]   Fast Multichannel Nonnegative Matrix Factorization With Directivity-Aware Jointly-Diagonalizable Spatial Covariance Matrices for Blind Source Separation [J].
Sekiguchi, Kouhei ;
Bando, Yoshiaki ;
Nugraha, Aditya Arie ;
Yoshii, Kazuyoshi ;
Kawahara, Tatsuya .
IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING, 2020, 28 :2610-2625
[22]   Underdetermined Blind Source Separation Combining Tensor Decomposition and Nonnegative Matrix Factorization [J].
Xie, Yuan ;
Xie, Kan ;
Yang, Junjie ;
Xie, Shengli .
SYMMETRY-BASEL, 2018, 10 (10)
[23]   Nonnegative signal factorization with learnt instrument models for sound source separation in close-microphone recordings [J].
Carabias-Orti, Julio J. ;
Cobos, Maximo ;
Vera-Candeas, Pedro ;
Rodriguez-Serrano, Francisco J. .
EURASIP JOURNAL ON ADVANCES IN SIGNAL PROCESSING, 2013,
[24]   Unsupervised Robust Speech Enhancement Based on Alpha-Stable Fast Multichannel Nonnegative Matrix Factorization [J].
Fontaine, Mathieu ;
Sekiguchi, Kouhei ;
Nugraha, Aditya Arie ;
Yoshii, Kazuyoshi .
INTERSPEECH 2020, 2020, :4541-4545
[25]   Hybrid Projective Nonnegative Matrix Factorization With Drum Dictionaries for Harmonic/Percussive Source Separation [J].
Laroche, Clement ;
Kowalski, Matthieu ;
Papadopoulos, Helene ;
Richard, Gael .
IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING, 2018, 26 (09) :1499-1511
[26]   Supervised Audio Source Separation Based on Nonnegative Matrix Factorization with Cosine Similarity Penalty [J].
Iwase, Yuta ;
Kitamura, Daichi .
IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES, 2022, E105A (06) :906-913
[27]   Monaural sound source separation by nonnegative matrix factorization with tempora continuity and sparseness criteria [J].
Virtanen, Tuomas .
IEEE TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING, 2007, 15 (03) :1066-1074
[28]   Notes on Nonnegative Tensor Factorization of the Spectrogram for Audio Source Separation: Statistical Insights and Towards Self-Clustering of the Spatial Cues [J].
Fevotte, Cedric ;
Ozerov, Alexey .
EXPLORING MUSIC CONTENTS, 2011, 6684 :102-+
[29]   On-Line Monaural Ambience Extraction Algorithm for Multichannel Audio Upmixing System Based on Nonnegative Matrix Factorization [J].
Lee, Seokjin ;
Pang, Hee-Suk .
IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES, 2015, E98A (01) :415-420
[30]   Blind separation of analytes in nuclear magnetic resonance spectroscopy: Improved model for nonnegative matrix factorization [J].
Kopriva, Ivica ;
Jeric, Ivanka .
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS, 2014, 137 :47-56
12345