Target Speech Signal Enhancement Based on Deep Neural Networks

Xin Zhang; Mingjiang Wang; Xiaoguang Xuan; Fengjiao Sun

doi:10.1109/ICICSP48821.2019.8958562

Target Speech Signal Enhancement Based on Deep Neural Networks

Xin Zhang
, Mingjiang Wang
, Xiaoguang Xuan
, Fengjiao Sun

Harbin Institute of Technology Shenzhen

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

This paper presents a speech enhancement system based on deep neural network (DNN) and covers the signal processing methods used in speech enhancement and the fundamentals of DNN. DNNs containing multiple hidden layers have the capability to suppress noise by learning the relationship between noisy speech and target clean speech. A series of experiments were carried out using the Chinese corpus to evaluate the performance of the speech enhancement model. Meanwhile, the model was identified as a result of a good generalization capability in mismatched noise types. In addition, the model is compared with OMLSA in terms of the quality of speech signals.

Original language	English
Title of host publication	2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	241-245
Number of pages	5
ISBN (Electronic)	9781728151021
DOIs	https://doi.org/10.1109/ICICSP48821.2019.8958562
State	Published - Sep 2019
Externally published	Yes
Event	2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019 - Weihai, China Duration: 28 Sep 2019 → 30 Sep 2019

Publication series

Name	2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019

Conference

Conference	2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019
Country/Territory	China
City	Weihai
Period	28/09/19 → 30/09/19

Keywords

Deep neural networks
Speech signal processing
Target speech signal enhancement

Access to Document

10.1109/ICICSP48821.2019.8958562

Cite this

Zhang, X., Wang, M., Xuan, X., & Sun, F. (2019). Target Speech Signal Enhancement Based on Deep Neural Networks. In 2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019 (pp. 241-245). Article 8958562 (2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICICSP48821.2019.8958562

Zhang, Xin ; Wang, Mingjiang ; Xuan, Xiaoguang et al. / Target Speech Signal Enhancement Based on Deep Neural Networks. 2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 241-245 (2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019).

@inproceedings{fae0befb4d0042b0a1e506194e9af05f,

title = "Target Speech Signal Enhancement Based on Deep Neural Networks",

abstract = "This paper presents a speech enhancement system based on deep neural network (DNN) and covers the signal processing methods used in speech enhancement and the fundamentals of DNN. DNNs containing multiple hidden layers have the capability to suppress noise by learning the relationship between noisy speech and target clean speech. A series of experiments were carried out using the Chinese corpus to evaluate the performance of the speech enhancement model. Meanwhile, the model was identified as a result of a good generalization capability in mismatched noise types. In addition, the model is compared with OMLSA in terms of the quality of speech signals.",

keywords = "Deep neural networks, Speech signal processing, Target speech signal enhancement",

author = "Xin Zhang and Mingjiang Wang and Xiaoguang Xuan and Fengjiao Sun",

note = "Publisher Copyright: {\textcopyright} 2019 IEEE.; 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019 ; Conference date: 28-09-2019 Through 30-09-2019",

year = "2019",

month = sep,

doi = "10.1109/ICICSP48821.2019.8958562",

language = "英语",

series = "2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "241--245",

booktitle = "2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019",

address = "美国",

}

Zhang, X, Wang, M, Xuan, X & Sun, F 2019, Target Speech Signal Enhancement Based on Deep Neural Networks. in 2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019., 8958562, 2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019, Institute of Electrical and Electronics Engineers Inc., pp. 241-245, 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019, Weihai, China, 28/09/19. https://doi.org/10.1109/ICICSP48821.2019.8958562

Target Speech Signal Enhancement Based on Deep Neural Networks. / Zhang, Xin; Wang, Mingjiang; Xuan, Xiaoguang et al.
2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019. Institute of Electrical and Electronics Engineers Inc., 2019. p. 241-245 8958562 (2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Target Speech Signal Enhancement Based on Deep Neural Networks

AU - Zhang, Xin

AU - Wang, Mingjiang

AU - Xuan, Xiaoguang

AU - Sun, Fengjiao

PY - 2019/9

Y1 - 2019/9

N2 - This paper presents a speech enhancement system based on deep neural network (DNN) and covers the signal processing methods used in speech enhancement and the fundamentals of DNN. DNNs containing multiple hidden layers have the capability to suppress noise by learning the relationship between noisy speech and target clean speech. A series of experiments were carried out using the Chinese corpus to evaluate the performance of the speech enhancement model. Meanwhile, the model was identified as a result of a good generalization capability in mismatched noise types. In addition, the model is compared with OMLSA in terms of the quality of speech signals.

AB - This paper presents a speech enhancement system based on deep neural network (DNN) and covers the signal processing methods used in speech enhancement and the fundamentals of DNN. DNNs containing multiple hidden layers have the capability to suppress noise by learning the relationship between noisy speech and target clean speech. A series of experiments were carried out using the Chinese corpus to evaluate the performance of the speech enhancement model. Meanwhile, the model was identified as a result of a good generalization capability in mismatched noise types. In addition, the model is compared with OMLSA in terms of the quality of speech signals.

KW - Deep neural networks

KW - Speech signal processing

KW - Target speech signal enhancement

UR - https://www.scopus.com/pages/publications/85078925275

U2 - 10.1109/ICICSP48821.2019.8958562

DO - 10.1109/ICICSP48821.2019.8958562

M3 - 会议稿件

AN - SCOPUS:85078925275

T3 - 2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019

SP - 241

EP - 245

BT - 2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019

Y2 - 28 September 2019 through 30 September 2019

ER -

Zhang X, Wang M, Xuan X, Sun F. Target Speech Signal Enhancement Based on Deep Neural Networks. In 2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019. Institute of Electrical and Electronics Engineers Inc. 2019. p. 241-245. 8958562. (2019 2nd IEEE International Conference on Information Communication and Signal Processing, ICICSP 2019). doi: 10.1109/ICICSP48821.2019.8958562

Target Speech Signal Enhancement Based on Deep Neural Networks

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this