Noise-Robust Voice Conversion Based on Sparse Spectral Mapping Using Non-negative Matrix Factorization

Ryo AIHARA; Ryoichi TAKASHIMA; Tetsuya TAKIGUCHI; Yasuo ARIKI

doi:10.1587/transinf.E97.D.1411

Noise-Robust Voice Conversion Based on Sparse Spectral Mapping Using Non-negative Matrix Factorization

Ryo AIHARA, Ryoichi TAKASHIMA, Tetsuya TAKIGUCHI, Yasuo ARIKI

Full Text Views

0

Cite this

Summary :

This paper presents a voice conversion (VC) technique for noisy environments based on a sparse representation of speech. Sparse representation-based VC using Non-negative matrix factorization (NMF) is employed for noise-added spectral conversion between different speakers. In our previous exemplar-based VC method, source exemplars and target exemplars are extracted from parallel training data, having the same texts uttered by the source and target speakers. The input source signal is represented using the source exemplars and their weights. Then, the converted speech is constructed from the target exemplars and the weights related to the source exemplars. However, this exemplar-based approach needs to hold all training exemplars (frames), and it requires high computation times to obtain the weights of the source exemplars. In this paper, we propose a framework to train the basis matrices of the source and target exemplars so that they have a common weight matrix. By using the basis matrices instead of the exemplars, the VC is performed with lower computation times than with the exemplar-based method. The effectiveness of this method was confirmed by comparing its effectiveness (in speaker conversion experiments using noise-added speech data) with that of an exemplar-based method and a conventional Gaussian mixture model (GMM)-based method.

Publication: IEICE TRANSACTIONS on Information Vol.E97-D No.6 pp.1411-1418

Publication Date: 2014/06/01

Publicized

Online ISSN: 1745-1361

DOI: 10.1587/transinf.E97.D.1411

Type of Manuscript: Special Section PAPER (Special Section on Advances in Modeling for Real-world Speech Information Processing and its Application)

Category: Voice Conversion and Speech Enhancement

Authors

Ryo AIHARA
  Kobe University
Ryoichi TAKASHIMA
  Kobe University
Tetsuya TAKIGUCHI
  Kobe University
Yasuo ARIKI
  Kobe University

Keyword

voice conversion, sparse representation, non-negative matrix factorization, noise robustness

Cite this

Copy

Ryo AIHARA, Ryoichi TAKASHIMA, Tetsuya TAKIGUCHI, Yasuo ARIKI, "Noise-Robust Voice Conversion Based on Sparse Spectral Mapping Using Non-negative Matrix Factorization" in IEICE TRANSACTIONS on Information, vol. E97-D, no. 6, pp. 1411-1418, June 2014, doi: 10.1587/transinf.E97.D.1411.
Abstract: This paper presents a voice conversion (VC) technique for noisy environments based on a sparse representation of speech. Sparse representation-based VC using Non-negative matrix factorization (NMF) is employed for noise-added spectral conversion between different speakers. In our previous exemplar-based VC method, source exemplars and target exemplars are extracted from parallel training data, having the same texts uttered by the source and target speakers. The input source signal is represented using the source exemplars and their weights. Then, the converted speech is constructed from the target exemplars and the weights related to the source exemplars. However, this exemplar-based approach needs to hold all training exemplars (frames), and it requires high computation times to obtain the weights of the source exemplars. In this paper, we propose a framework to train the basis matrices of the source and target exemplars so that they have a common weight matrix. By using the basis matrices instead of the exemplars, the VC is performed with lower computation times than with the exemplar-based method. The effectiveness of this method was confirmed by comparing its effectiveness (in speaker conversion experiments using noise-added speech data) with that of an exemplar-based method and a conventional Gaussian mixture model (GMM)-based method.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.E97.D.1411/_p

Copy

@ARTICLE{e97-d_6_1411,
author={Ryo AIHARA, Ryoichi TAKASHIMA, Tetsuya TAKIGUCHI, Yasuo ARIKI, },
journal={IEICE TRANSACTIONS on Information},
title={Noise-Robust Voice Conversion Based on Sparse Spectral Mapping Using Non-negative Matrix Factorization},
year={2014},
volume={E97-D},
number={6},
pages={1411-1418},
abstract={This paper presents a voice conversion (VC) technique for noisy environments based on a sparse representation of speech. Sparse representation-based VC using Non-negative matrix factorization (NMF) is employed for noise-added spectral conversion between different speakers. In our previous exemplar-based VC method, source exemplars and target exemplars are extracted from parallel training data, having the same texts uttered by the source and target speakers. The input source signal is represented using the source exemplars and their weights. Then, the converted speech is constructed from the target exemplars and the weights related to the source exemplars. However, this exemplar-based approach needs to hold all training exemplars (frames), and it requires high computation times to obtain the weights of the source exemplars. In this paper, we propose a framework to train the basis matrices of the source and target exemplars so that they have a common weight matrix. By using the basis matrices instead of the exemplars, the VC is performed with lower computation times than with the exemplar-based method. The effectiveness of this method was confirmed by comparing its effectiveness (in speaker conversion experiments using noise-added speech data) with that of an exemplar-based method and a conventional Gaussian mixture model (GMM)-based method.},
keywords={},
doi={10.1587/transinf.E97.D.1411},
ISSN={1745-1361},
month={June},}

Copy

TY - JOUR
TI - Noise-Robust Voice Conversion Based on Sparse Spectral Mapping Using Non-negative Matrix Factorization
T2 - IEICE TRANSACTIONS on Information
SP - 1411
EP - 1418
AU - Ryo AIHARA
AU - Ryoichi TAKASHIMA
AU - Tetsuya TAKIGUCHI
AU - Yasuo ARIKI
PY - 2014
DO - 10.1587/transinf.E97.D.1411
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E97-D
IS - 6
JA - IEICE TRANSACTIONS on Information
Y1 - June 2014
AB - This paper presents a voice conversion (VC) technique for noisy environments based on a sparse representation of speech. Sparse representation-based VC using Non-negative matrix factorization (NMF) is employed for noise-added spectral conversion between different speakers. In our previous exemplar-based VC method, source exemplars and target exemplars are extracted from parallel training data, having the same texts uttered by the source and target speakers. The input source signal is represented using the source exemplars and their weights. Then, the converted speech is constructed from the target exemplars and the weights related to the source exemplars. However, this exemplar-based approach needs to hold all training exemplars (frames), and it requires high computation times to obtain the weights of the source exemplars. In this paper, we propose a framework to train the basis matrices of the source and target exemplars so that they have a common weight matrix. By using the basis matrices instead of the exemplars, the VC is performed with lower computation times than with the exemplar-based method. The effectiveness of this method was confirmed by comparing its effectiveness (in speaker conversion experiments using noise-added speech data) with that of an exemplar-based method and a conventional Gaussian mixture model (GMM)-based method.
ER -