Denoi-SpEx+: A Speaker Extraction Network based Speech Dialogue System

Yun Hao; Xiangkang Huang; Huichou Huang; Qingyao Wu

doi:10.1109/ICEBE52470.2021.00030

Denoi-SpEx+: A Speaker Extraction Network based Speech Dialogue System

Yun Hao
, Xiangkang Huang
, Huichou Huang^*
, Qingyao Wu^*

^*Corresponding author for this work

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

2 Citations (Scopus)

Abstract

The speech dialogue system has gradually been widely used in daily life. Users can consult and communicate with the system through natural language. However, in practical applications, third-person background sounds and background noise interference in real dialogue scenes will be encountered. The uncertainty and complexity of these background sounds will have a bad impact on the recognition of the system. A good speech enhancement module can help us to separate the target speaker from the original speech. Recently, a solution called SpEx+ was proposed from the time domain, but SpEx+ needs a reference speech to assist in training. This reference speech may have noise in actual applications that will affect performance. Therefore, we propose a Denoi-SpEx+ model. Before the reference speech is input to the network, a speech denoising network is added, so that the quality of speech separation in practical applications can be guaranteed. Experiments show that our model can significantly improve the performance of speech separation model of noisy reference speech. ©2021 IEEE.

Original language	English
Title of host publication	Proceedings - 2021 IEEE International Conference on e-Business Engineering, ICEBE 2021
Place of Publication	Los Alamitos, Calif.
Publisher	IEEE
Pages	49-53
ISBN (Electronic)	978-1-6654-4418-7
ISBN (Print)	978-1-6654-4419-4
DOIs	https://doi.org/10.1109/ICEBE52470.2021.00030
Publication status	Published - Nov 2021
Externally published	Yes
Event	17th IEEE International Conference on e-Business Engineering (ICEBE 2021) - Guangzhou, China Duration: 12 Nov 2021 → 14 Nov 2021

Publication series

Name	International Conference on e-Business Engineering

Conference

Conference	17th IEEE International Conference on e-Business Engineering (ICEBE 2021)
Place	China
City	Guangzhou
Period	12/11/21 → 14/11/21

Funding

This work was supported by National Natural Science Foundation of China (NSFC) 61876208, Key-Area Research and Development Program of Guangdong Province 2018B010108002

Research Keywords

speech dialogue system
speech separation
Denoi-SpEx+
SEPARATION
ENHANCEMENT

Access Output

10.1109/ICEBE52470.2021.00030

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

@inproceedings{0547ff1e6e4047579b6e2a74ccdcd1dc,

title = "Denoi-SpEx+: A Speaker Extraction Network based Speech Dialogue System",

abstract = "The speech dialogue system has gradually been widely used in daily life. Users can consult and communicate with the system through natural language. However, in practical applications, third-person background sounds and background noise interference in real dialogue scenes will be encountered. The uncertainty and complexity of these background sounds will have a bad impact on the recognition of the system. A good speech enhancement module can help us to separate the target speaker from the original speech. Recently, a solution called SpEx+ was proposed from the time domain, but SpEx+ needs a reference speech to assist in training. This reference speech may have noise in actual applications that will affect performance. Therefore, we propose a Denoi-SpEx+ model. Before the reference speech is input to the network, a speech denoising network is added, so that the quality of speech separation in practical applications can be guaranteed. Experiments show that our model can significantly improve the performance of speech separation model of noisy reference speech. {\textcopyright}2021 IEEE.",

keywords = "speech dialogue system, speech separation, Denoi-SpEx+, SEPARATION, ENHANCEMENT",

author = "Yun Hao and Xiangkang Huang and Huichou Huang and Qingyao Wu",

year = "2021",

month = nov,

doi = "10.1109/ICEBE52470.2021.00030",

language = "English",

isbn = "978-1-6654-4419-4",

series = "International Conference on e-Business Engineering",

publisher = "IEEE",

pages = "49--53",

booktitle = "Proceedings - 2021 IEEE International Conference on e-Business Engineering, ICEBE 2021",

address = "United States",

note = "17th IEEE International Conference on e-Business Engineering (ICEBE 2021) ; Conference date: 12-11-2021 Through 14-11-2021",

}

Hao, Y, Huang, X, Huang, H & Wu, Q 2021, Denoi-SpEx+: A Speaker Extraction Network based Speech Dialogue System. in Proceedings - 2021 IEEE International Conference on e-Business Engineering, ICEBE 2021. International Conference on e-Business Engineering, IEEE, Los Alamitos, Calif., pp. 49-53, 17th IEEE International Conference on e-Business Engineering (ICEBE 2021), Guangzhou, China, 12/11/21. https://doi.org/10.1109/ICEBE52470.2021.00030

Denoi-SpEx+: A Speaker Extraction Network based Speech Dialogue System. / Hao, Yun; Huang, Xiangkang; Huang, Huichou et al.
Proceedings - 2021 IEEE International Conference on e-Business Engineering, ICEBE 2021. Los Alamitos, Calif.: IEEE, 2021. p. 49-53 (International Conference on e-Business Engineering).

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

TY - GEN

T1 - Denoi-SpEx+

T2 - 17th IEEE International Conference on e-Business Engineering (ICEBE 2021)

AU - Hao, Yun

AU - Huang, Xiangkang

AU - Huang, Huichou

AU - Wu, Qingyao

PY - 2021/11

Y1 - 2021/11

N2 - The speech dialogue system has gradually been widely used in daily life. Users can consult and communicate with the system through natural language. However, in practical applications, third-person background sounds and background noise interference in real dialogue scenes will be encountered. The uncertainty and complexity of these background sounds will have a bad impact on the recognition of the system. A good speech enhancement module can help us to separate the target speaker from the original speech. Recently, a solution called SpEx+ was proposed from the time domain, but SpEx+ needs a reference speech to assist in training. This reference speech may have noise in actual applications that will affect performance. Therefore, we propose a Denoi-SpEx+ model. Before the reference speech is input to the network, a speech denoising network is added, so that the quality of speech separation in practical applications can be guaranteed. Experiments show that our model can significantly improve the performance of speech separation model of noisy reference speech. ©2021 IEEE.

AB - The speech dialogue system has gradually been widely used in daily life. Users can consult and communicate with the system through natural language. However, in practical applications, third-person background sounds and background noise interference in real dialogue scenes will be encountered. The uncertainty and complexity of these background sounds will have a bad impact on the recognition of the system. A good speech enhancement module can help us to separate the target speaker from the original speech. Recently, a solution called SpEx+ was proposed from the time domain, but SpEx+ needs a reference speech to assist in training. This reference speech may have noise in actual applications that will affect performance. Therefore, we propose a Denoi-SpEx+ model. Before the reference speech is input to the network, a speech denoising network is added, so that the quality of speech separation in practical applications can be guaranteed. Experiments show that our model can significantly improve the performance of speech separation model of noisy reference speech. ©2021 IEEE.

KW - speech dialogue system

KW - speech separation

KW - Denoi-SpEx+

KW - SEPARATION

KW - ENHANCEMENT

UR - http://gateway.isiknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcAuth=LinksAMR&SrcApp=PARTNER_APP&DestLinkType=FullRecord&DestApp=WOS&KeyUT=000942017800008

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85128720273&origin=recordpage

UR - https://www.scopus.com/pages/publications/85128720273

U2 - 10.1109/ICEBE52470.2021.00030

DO - 10.1109/ICEBE52470.2021.00030

M3 - RGC 32 - Refereed conference paper (with host publication)

SN - 978-1-6654-4419-4

T3 - International Conference on e-Business Engineering

SP - 49

EP - 53

BT - Proceedings - 2021 IEEE International Conference on e-Business Engineering, ICEBE 2021

PB - IEEE

CY - Los Alamitos, Calif.

Y2 - 12 November 2021 through 14 November 2021

ER -

Denoi-SpEx+: A Speaker Extraction Network based Speech Dialogue System

Abstract

Publication series

Conference

Funding

Research Keywords

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this