Deep convolutional neural network with mixup for environmental sound classification

Zhichao Zhang; Shugong Xu; Shan Cao; Shunqing Zhang

doi:10.1007/978-3-030-03335-4_31

Deep convolutional neural network with mixup for environmental sound classification

Zhichao Zhang, Shugong Xu^*, Shan Cao, Shunqing Zhang

^*Corresponding author for this work

Shanghai University

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

96 Citations (Scopus)

Abstract

Environmental sound classification (ESC) is an important and challenging problem. In contrast to speech, sound events have noise-like nature and may be produced by a wide variety of sources. In this paper, we propose to use a novel deep convolutional neural network for ESC tasks. Our network architecture uses stacked convolutional and pooling layers to extract high-level feature representations from spectrogram-like features. Furthermore, we apply mixup to ESC tasks and explore its impacts on classification performance and feature distribution. Experiments were conducted on UrbanSound8K, ESC-50 and ESC-10 datasets. Our experimental results demonstrated that our ESC system has achieved the state-of-the-art performance (83.7%) on UrbanSound8K and competitive performance on ESC-50 and ESC-10.

Original language	English
Title of host publication	Pattern Recognition and Computer Vision - First Chinese Conference, PRCV 2018, Proceedings
Editors	Cheng-Lin Liu, Tieniu Tan, Jie Zhou, Jian-Huang Lai, Xilin Chen, Nanning Zheng, Hongbin Zha
Publisher	Springer Verlag
Pages	356-367
Number of pages	12
ISBN (Print)	9783030033347
DOIs	https://doi.org/10.1007/978-3-030-03335-4_31
Publication status	Published - 2018
Externally published	Yes
Event	1st Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2018 - Guangzhou, China Duration: 23 Nov 2018 → 26 Nov 2018

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11257 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	1st Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2018
Country/Territory	China
City	Guangzhou
Period	23/11/18 → 26/11/18

Keywords

Convolutional neural network
Environmental sound classification
Mixup

Access to Document

10.1007/978-3-030-03335-4_31

Cite this

Zhang, Z., Xu, S., Cao, S., & Zhang, S. (2018). Deep convolutional neural network with mixup for environmental sound classification. In C.-L. Liu, T. Tan, J. Zhou, J.-H. Lai, X. Chen, N. Zheng, & H. Zha (Eds.), Pattern Recognition and Computer Vision - First Chinese Conference, PRCV 2018, Proceedings (pp. 356-367). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11257 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-03335-4_31

Zhang, Zhichao ; Xu, Shugong ; Cao, Shan et al. / Deep convolutional neural network with mixup for environmental sound classification. Pattern Recognition and Computer Vision - First Chinese Conference, PRCV 2018, Proceedings. editor / Cheng-Lin Liu ; Tieniu Tan ; Jie Zhou ; Jian-Huang Lai ; Xilin Chen ; Nanning Zheng ; Hongbin Zha. Springer Verlag, 2018. pp. 356-367 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{caa58e9c0a4b452982bdad5d8944c25b,

title = "Deep convolutional neural network with mixup for environmental sound classification",

abstract = "Environmental sound classification (ESC) is an important and challenging problem. In contrast to speech, sound events have noise-like nature and may be produced by a wide variety of sources. In this paper, we propose to use a novel deep convolutional neural network for ESC tasks. Our network architecture uses stacked convolutional and pooling layers to extract high-level feature representations from spectrogram-like features. Furthermore, we apply mixup to ESC tasks and explore its impacts on classification performance and feature distribution. Experiments were conducted on UrbanSound8K, ESC-50 and ESC-10 datasets. Our experimental results demonstrated that our ESC system has achieved the state-of-the-art performance (83.7%) on UrbanSound8K and competitive performance on ESC-50 and ESC-10.",

keywords = "Convolutional neural network, Environmental sound classification, Mixup",

author = "Zhichao Zhang and Shugong Xu and Shan Cao and Shunqing Zhang",

note = "Publisher Copyright: {\textcopyright} Springer Nature Switzerland AG 2018.; 1st Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2018 ; Conference date: 23-11-2018 Through 26-11-2018",

year = "2018",

doi = "10.1007/978-3-030-03335-4_31",

language = "English",

isbn = "9783030033347",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "356--367",

editor = "Cheng-Lin Liu and Tieniu Tan and Jie Zhou and Jian-Huang Lai and Xilin Chen and Nanning Zheng and Hongbin Zha",

booktitle = "Pattern Recognition and Computer Vision - First Chinese Conference, PRCV 2018, Proceedings",

}

Zhang, Z, Xu, S, Cao, S & Zhang, S 2018, Deep convolutional neural network with mixup for environmental sound classification. in C-L Liu, T Tan, J Zhou, J-H Lai, X Chen, N Zheng & H Zha (eds), Pattern Recognition and Computer Vision - First Chinese Conference, PRCV 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11257 LNCS, Springer Verlag, pp. 356-367, 1st Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2018, Guangzhou, China, 23/11/18. https://doi.org/10.1007/978-3-030-03335-4_31

Deep convolutional neural network with mixup for environmental sound classification. / Zhang, Zhichao; Xu, Shugong; Cao, Shan et al.
Pattern Recognition and Computer Vision - First Chinese Conference, PRCV 2018, Proceedings. ed. / Cheng-Lin Liu; Tieniu Tan; Jie Zhou; Jian-Huang Lai; Xilin Chen; Nanning Zheng; Hongbin Zha. Springer Verlag, 2018. p. 356-367 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11257 LNCS).

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

TY - GEN

T1 - Deep convolutional neural network with mixup for environmental sound classification

AU - Zhang, Zhichao

AU - Xu, Shugong

AU - Cao, Shan

AU - Zhang, Shunqing

N1 - Publisher Copyright: © Springer Nature Switzerland AG 2018.

PY - 2018

Y1 - 2018

N2 - Environmental sound classification (ESC) is an important and challenging problem. In contrast to speech, sound events have noise-like nature and may be produced by a wide variety of sources. In this paper, we propose to use a novel deep convolutional neural network for ESC tasks. Our network architecture uses stacked convolutional and pooling layers to extract high-level feature representations from spectrogram-like features. Furthermore, we apply mixup to ESC tasks and explore its impacts on classification performance and feature distribution. Experiments were conducted on UrbanSound8K, ESC-50 and ESC-10 datasets. Our experimental results demonstrated that our ESC system has achieved the state-of-the-art performance (83.7%) on UrbanSound8K and competitive performance on ESC-50 and ESC-10.

AB - Environmental sound classification (ESC) is an important and challenging problem. In contrast to speech, sound events have noise-like nature and may be produced by a wide variety of sources. In this paper, we propose to use a novel deep convolutional neural network for ESC tasks. Our network architecture uses stacked convolutional and pooling layers to extract high-level feature representations from spectrogram-like features. Furthermore, we apply mixup to ESC tasks and explore its impacts on classification performance and feature distribution. Experiments were conducted on UrbanSound8K, ESC-50 and ESC-10 datasets. Our experimental results demonstrated that our ESC system has achieved the state-of-the-art performance (83.7%) on UrbanSound8K and competitive performance on ESC-50 and ESC-10.

KW - Convolutional neural network

KW - Environmental sound classification

KW - Mixup

UR - http://www.scopus.com/inward/record.url?scp=85057156514&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-03335-4_31

DO - 10.1007/978-3-030-03335-4_31

M3 - Conference Proceeding

AN - SCOPUS:85057156514

SN - 9783030033347

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 356

EP - 367

BT - Pattern Recognition and Computer Vision - First Chinese Conference, PRCV 2018, Proceedings

A2 - Liu, Cheng-Lin

A2 - Tan, Tieniu

A2 - Zhou, Jie

A2 - Lai, Jian-Huang

A2 - Chen, Xilin

A2 - Zheng, Nanning

A2 - Zha, Hongbin

PB - Springer Verlag

T2 - 1st Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2018

Y2 - 23 November 2018 through 26 November 2018

ER -

Zhang Z, Xu S, Cao S, Zhang S. Deep convolutional neural network with mixup for environmental sound classification. In Liu CL, Tan T, Zhou J, Lai JH, Chen X, Zheng N, Zha H, editors, Pattern Recognition and Computer Vision - First Chinese Conference, PRCV 2018, Proceedings. Springer Verlag. 2018. p. 356-367. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-03335-4_31

Deep convolutional neural network with mixup for environmental sound classification

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this