MANNER: MULTI-VIEW ATTENTION NETWORK FOR NOISE ERASURE

Hyun Joon Park; Byung Ha Kang; Wooseok Shin; Jin Sob Kim; Sung Won Han

doi:10.1109/ICASSP43922.2022.9747120

MANNER: MULTI-VIEW ATTENTION NETWORK FOR NOISE ERASURE

Hyun Joon Park, Byung Ha Kang, Wooseok Shin, Jin Sob Kim, Sung Won Han

School of Industrial Management Engineering

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

1 Citation (Scopus)

Abstract

In the field of speech enhancement, time domain methods have difficulties in achieving both high performance and efficiency. Recently, dual-path models have been adopted to represent long sequential features, but they still have limited representations and poor memory efficiency. In this study, we propose Multi-view Attention Network for Noise ERasure (MANNER) consisting of a convolutional encoder-decoder with a multi-view attention block, applied to the time-domain signals. MANNER efficiently extracts three different representations from noisy speech and estimates high-quality clean speech. We evaluated MANNER on the VoiceBank-DEMAND dataset in terms of five objective speech quality metrics. Experimental results show that MANNER achieves state-of-the-art performance while efficiently processing noisy speech.

Original language	English
Title of host publication	2022 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 - Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	7842-7846
Number of pages	5
ISBN (Electronic)	9781665405409
DOIs	https://doi.org/10.1109/ICASSP43922.2022.9747120
Publication status	Published - 2022
Event	47th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 - Virtual, Online, Singapore Duration: 2022 May 23 → 2022 May 27

Publication series

Name	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Volume	2022-May
ISSN (Print)	1520-6149

Conference

Conference	47th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022
Country/Territory	Singapore
City	Virtual, Online
Period	22/5/23 → 22/5/27

Keywords

multi-view attention
speech enhancement
time domain
u-net

ASJC Scopus subject areas

Software
Signal Processing
Electrical and Electronic Engineering

Access to Document

10.1109/ICASSP43922.2022.9747120

Cite this

Park, H. J., Kang, B. H., Shin, W., Kim, J. S., & Han, S. W. (2022). MANNER: MULTI-VIEW ATTENTION NETWORK FOR NOISE ERASURE. In 2022 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 - Proceedings (pp. 7842-7846). (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2022-May). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICASSP43922.2022.9747120

MANNER : MULTI-VIEW ATTENTION NETWORK FOR NOISE ERASURE. / Park, Hyun Joon; Kang, Byung Ha; Shin, Wooseok et al.

2022 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2022. p. 7842-7846 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2022-May).

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

Park, HJ, Kang, BH, Shin, W, Kim, JS & Han, SW 2022, MANNER: MULTI-VIEW ATTENTION NETWORK FOR NOISE ERASURE. in 2022 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 - Proceedings. ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, vol. 2022-May, Institute of Electrical and Electronics Engineers Inc., pp. 7842-7846, 47th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022, Virtual, Online, Singapore, 22/5/23. https://doi.org/10.1109/ICASSP43922.2022.9747120

Park HJ, Kang BH, Shin W, Kim JS, Han SW. MANNER: MULTI-VIEW ATTENTION NETWORK FOR NOISE ERASURE. In 2022 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2022. p. 7842-7846. (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). doi: 10.1109/ICASSP43922.2022.9747120

Park, Hyun Joon ; Kang, Byung Ha ; Shin, Wooseok et al. / MANNER : MULTI-VIEW ATTENTION NETWORK FOR NOISE ERASURE. 2022 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2022. pp. 7842-7846 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

@inproceedings{ee8813ffd1c8446b8fbe05736be26966,

title = "MANNER: MULTI-VIEW ATTENTION NETWORK FOR NOISE ERASURE",

abstract = "In the field of speech enhancement, time domain methods have difficulties in achieving both high performance and efficiency. Recently, dual-path models have been adopted to represent long sequential features, but they still have limited representations and poor memory efficiency. In this study, we propose Multi-view Attention Network for Noise ERasure (MANNER) consisting of a convolutional encoder-decoder with a multi-view attention block, applied to the time-domain signals. MANNER efficiently extracts three different representations from noisy speech and estimates high-quality clean speech. We evaluated MANNER on the VoiceBank-DEMAND dataset in terms of five objective speech quality metrics. Experimental results show that MANNER achieves state-of-the-art performance while efficiently processing noisy speech.",

keywords = "multi-view attention, speech enhancement, time domain, u-net",

author = "Park, {Hyun Joon} and Kang, {Byung Ha} and Wooseok Shin and Kim, {Jin Sob} and Han, {Sung Won}",

note = "Funding Information: This research was supported by Brain Korea 21 FOUR. This research was also supported by Korea University Grant (K2107521) and a Korea Tech-noComplex Foundation Grant (R2112651). Funding Information: This research was supported by Brain Korea 21 FOUR. This research was also supported by Korea University Grant (K2107521) and a Korea TechnoComplex Foundation Grant (R2112651). Publisher Copyright: {\textcopyright} 2022 IEEE; 47th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 ; Conference date: 23-05-2022 Through 27-05-2022",

year = "2022",

doi = "10.1109/ICASSP43922.2022.9747120",

language = "English",

series = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

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

pages = "7842--7846",

booktitle = "2022 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 - Proceedings",

}

TY - GEN

T1 - MANNER

T2 - 47th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022

AU - Park, Hyun Joon

AU - Kang, Byung Ha

AU - Shin, Wooseok

AU - Kim, Jin Sob

AU - Han, Sung Won

N1 - Funding Information: This research was supported by Brain Korea 21 FOUR. This research was also supported by Korea University Grant (K2107521) and a Korea Tech-noComplex Foundation Grant (R2112651). Funding Information: This research was supported by Brain Korea 21 FOUR. This research was also supported by Korea University Grant (K2107521) and a Korea TechnoComplex Foundation Grant (R2112651). Publisher Copyright: © 2022 IEEE

PY - 2022

Y1 - 2022

N2 - In the field of speech enhancement, time domain methods have difficulties in achieving both high performance and efficiency. Recently, dual-path models have been adopted to represent long sequential features, but they still have limited representations and poor memory efficiency. In this study, we propose Multi-view Attention Network for Noise ERasure (MANNER) consisting of a convolutional encoder-decoder with a multi-view attention block, applied to the time-domain signals. MANNER efficiently extracts three different representations from noisy speech and estimates high-quality clean speech. We evaluated MANNER on the VoiceBank-DEMAND dataset in terms of five objective speech quality metrics. Experimental results show that MANNER achieves state-of-the-art performance while efficiently processing noisy speech.

AB - In the field of speech enhancement, time domain methods have difficulties in achieving both high performance and efficiency. Recently, dual-path models have been adopted to represent long sequential features, but they still have limited representations and poor memory efficiency. In this study, we propose Multi-view Attention Network for Noise ERasure (MANNER) consisting of a convolutional encoder-decoder with a multi-view attention block, applied to the time-domain signals. MANNER efficiently extracts three different representations from noisy speech and estimates high-quality clean speech. We evaluated MANNER on the VoiceBank-DEMAND dataset in terms of five objective speech quality metrics. Experimental results show that MANNER achieves state-of-the-art performance while efficiently processing noisy speech.

KW - multi-view attention

KW - speech enhancement

KW - time domain

KW - u-net

UR - http://www.scopus.com/inward/record.url?scp=85131230444&partnerID=8YFLogxK

U2 - 10.1109/ICASSP43922.2022.9747120

DO - 10.1109/ICASSP43922.2022.9747120

M3 - Conference contribution

AN - SCOPUS:85131230444

T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

SP - 7842

EP - 7846

BT - 2022 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 23 May 2022 through 27 May 2022

ER -

MANNER: MULTI-VIEW ATTENTION NETWORK FOR NOISE ERASURE

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this