Singing voice detection using multi-feature deep fusion with CNN

Xulong Zhang; Shengchen Li; Zijin Li; Shizhe Chen; Yongwei Gao; Wei Li

doi:10.1007/978-981-15-2756-2_4

Singing voice detection using multi-feature deep fusion with CNN

Xulong Zhang, Shengchen Li, Zijin Li, Shizhe Chen, Yongwei Gao, Wei Li^*

^*Corresponding author for this work

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

6 Citations (Scopus)

Abstract

The problem of singing voice detection is to segment a song into vocal and non-vocal parts. Commonly used methods usually train a model on a set of frame-based features and then predict the unknown frames by the model. However, the multi-dimensional features are usually concatenated together for each frame, with little consideration of spatial information. Hence, a deep fusion method of the Multi-feature dimensions with Convolution Neural Networks (CNN) is proposed. A one dimension convolution is made on feature dimensions for each frames, then the high-level features obtained can be used for a direct binary classification. The performance of the proposed method is on par with the state-of-art methods on public dataset.

Original language	English
Title of host publication	Proceedings of the 7th Conference on Sound and Music Technology CSMT 2019, Revised Selected Papers
Editors	Haifeng Li, Lin Ma, Shengchen Li, Chunying Fang, Yidan Zhu
Publisher	Springer
Pages	41-52
Number of pages	12
ISBN (Print)	9789811527555
DOIs	https://doi.org/10.1007/978-981-15-2756-2_4
Publication status	Published - 2020
Externally published	Yes
Event	7th Conference on Sound and Music Technology, CSMT 2019 - Harbin, China Duration: 26 Dec 2019 → 29 Dec 2019

Publication series

Name	Lecture Notes in Electrical Engineering
Volume	635
ISSN (Print)	1876-1100
ISSN (Electronic)	1876-1119

Conference

Conference	7th Conference on Sound and Music Technology, CSMT 2019
Country/Territory	China
City	Harbin
Period	26/12/19 → 29/12/19

Keywords

Convolution neural network (CNN)
Deep learning
Multi-feature fusion
Singing voice detection (SVD)

Access to Document

10.1007/978-981-15-2756-2_4

Cite this

Zhang, X., Li, S., Li, Z., Chen, S., Gao, Y., & Li, W. (2020). Singing voice detection using multi-feature deep fusion with CNN. In H. Li, L. Ma, S. Li, C. Fang, & Y. Zhu (Eds.), Proceedings of the 7th Conference on Sound and Music Technology CSMT 2019, Revised Selected Papers (pp. 41-52). (Lecture Notes in Electrical Engineering; Vol. 635). Springer. https://doi.org/10.1007/978-981-15-2756-2_4

@inproceedings{3bb3b0d9a50142d6826cd28ef604925f,

title = "Singing voice detection using multi-feature deep fusion with CNN",

abstract = "The problem of singing voice detection is to segment a song into vocal and non-vocal parts. Commonly used methods usually train a model on a set of frame-based features and then predict the unknown frames by the model. However, the multi-dimensional features are usually concatenated together for each frame, with little consideration of spatial information. Hence, a deep fusion method of the Multi-feature dimensions with Convolution Neural Networks (CNN) is proposed. A one dimension convolution is made on feature dimensions for each frames, then the high-level features obtained can be used for a direct binary classification. The performance of the proposed method is on par with the state-of-art methods on public dataset.",

keywords = "Convolution neural network (CNN), Deep learning, Multi-feature fusion, Singing voice detection (SVD)",

author = "Xulong Zhang and Shengchen Li and Zijin Li and Shizhe Chen and Yongwei Gao and Wei Li",

note = "Publisher Copyright: {\textcopyright} Springer Nature Singapore Pte Ltd 2020.; 7th Conference on Sound and Music Technology, CSMT 2019 ; Conference date: 26-12-2019 Through 29-12-2019",

year = "2020",

doi = "10.1007/978-981-15-2756-2_4",

language = "English",

isbn = "9789811527555",

series = "Lecture Notes in Electrical Engineering",

publisher = "Springer",

pages = "41--52",

editor = "Haifeng Li and Lin Ma and Shengchen Li and Chunying Fang and Yidan Zhu",

booktitle = "Proceedings of the 7th Conference on Sound and Music Technology CSMT 2019, Revised Selected Papers",

}

Zhang, X, Li, S, Li, Z, Chen, S, Gao, Y & Li, W 2020, Singing voice detection using multi-feature deep fusion with CNN. in H Li, L Ma, S Li, C Fang & Y Zhu (eds), Proceedings of the 7th Conference on Sound and Music Technology CSMT 2019, Revised Selected Papers. Lecture Notes in Electrical Engineering, vol. 635, Springer, pp. 41-52, 7th Conference on Sound and Music Technology, CSMT 2019, Harbin, China, 26/12/19. https://doi.org/10.1007/978-981-15-2756-2_4

Singing voice detection using multi-feature deep fusion with CNN. / Zhang, Xulong; Li, Shengchen; Li, Zijin et al.
Proceedings of the 7th Conference on Sound and Music Technology CSMT 2019, Revised Selected Papers. ed. / Haifeng Li; Lin Ma; Shengchen Li; Chunying Fang; Yidan Zhu. Springer, 2020. p. 41-52 (Lecture Notes in Electrical Engineering; Vol. 635).

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

TY - GEN

T1 - Singing voice detection using multi-feature deep fusion with CNN

AU - Zhang, Xulong

AU - Li, Shengchen

AU - Li, Zijin

AU - Chen, Shizhe

AU - Gao, Yongwei

AU - Li, Wei

N1 - Publisher Copyright: © Springer Nature Singapore Pte Ltd 2020.

PY - 2020

Y1 - 2020

N2 - The problem of singing voice detection is to segment a song into vocal and non-vocal parts. Commonly used methods usually train a model on a set of frame-based features and then predict the unknown frames by the model. However, the multi-dimensional features are usually concatenated together for each frame, with little consideration of spatial information. Hence, a deep fusion method of the Multi-feature dimensions with Convolution Neural Networks (CNN) is proposed. A one dimension convolution is made on feature dimensions for each frames, then the high-level features obtained can be used for a direct binary classification. The performance of the proposed method is on par with the state-of-art methods on public dataset.

AB - The problem of singing voice detection is to segment a song into vocal and non-vocal parts. Commonly used methods usually train a model on a set of frame-based features and then predict the unknown frames by the model. However, the multi-dimensional features are usually concatenated together for each frame, with little consideration of spatial information. Hence, a deep fusion method of the Multi-feature dimensions with Convolution Neural Networks (CNN) is proposed. A one dimension convolution is made on feature dimensions for each frames, then the high-level features obtained can be used for a direct binary classification. The performance of the proposed method is on par with the state-of-art methods on public dataset.

KW - Convolution neural network (CNN)

KW - Deep learning

KW - Multi-feature fusion

KW - Singing voice detection (SVD)

UR - http://www.scopus.com/inward/record.url?scp=85078450640&partnerID=8YFLogxK

U2 - 10.1007/978-981-15-2756-2_4

DO - 10.1007/978-981-15-2756-2_4

M3 - Conference Proceeding

AN - SCOPUS:85078450640

SN - 9789811527555

T3 - Lecture Notes in Electrical Engineering

SP - 41

EP - 52

BT - Proceedings of the 7th Conference on Sound and Music Technology CSMT 2019, Revised Selected Papers

A2 - Li, Haifeng

A2 - Ma, Lin

A2 - Li, Shengchen

A2 - Fang, Chunying

A2 - Zhu, Yidan

PB - Springer

T2 - 7th Conference on Sound and Music Technology, CSMT 2019

Y2 - 26 December 2019 through 29 December 2019

ER -

Singing voice detection using multi-feature deep fusion with CNN

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this