OveNet: A Hyper-Range U-Net for Singing Voice Separation

Chi Sheng Wu; Shiang Lee; Von Wun Soo

doi:10.1109/ISM46123.2019.00034

OveNet: A Hyper-Range U-Net for Singing Voice Separation

Chi Sheng Wu, Shiang Lee, Von Wun Soo

National Tsing Hua University

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

Abstract

In the audio source separation topic, most researchers based on deep learning methods ignored higher-frequency signals due to lack of efficient data compression method. We propose a new model named OvertoneNet (OveNet) that adopts two novel concepts, frequency 1x1 convolution layers, and complex-spectrogram channels, to handle the 44.1k audio signals (Hi-Res audio signals) containing full overtones. The result shows that OveNet performs well in both objective and subjective evaluation on interference using limited training data from SiSEC2018.

Original language	English
Title of host publication	Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	148-151
Number of pages	4
ISBN (Electronic)	9781728156064
DOIs	https://doi.org/10.1109/ISM46123.2019.00034
State	Published - 12 2019
Externally published	Yes
Event	21st IEEE International Symposium on Multimedia, ISM 2019 - San Diego, United States Duration: 09 12 2019 → 11 12 2019

Publication series

Name	Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019

Conference

Conference	21st IEEE International Symposium on Multimedia, ISM 2019
Country/Territory	United States
City	San Diego
Period	09/12/19 → 11/12/19

Bibliographical note

Publisher Copyright:
© 2019 IEEE.

Keywords

U-Net
audio source separation
convolutional neural networks

Access to Document

10.1109/ISM46123.2019.00034

Cite this

Wu, C. S., Lee, S., & Soo, V. W. (2019). OveNet: A Hyper-Range U-Net for Singing Voice Separation. In Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019 (pp. 148-151). Article 8959050 (Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ISM46123.2019.00034

@inproceedings{9f98a0f2e4e246a6bca58d57f811bf83,

title = "OveNet: A Hyper-Range U-Net for Singing Voice Separation",

abstract = "In the audio source separation topic, most researchers based on deep learning methods ignored higher-frequency signals due to lack of efficient data compression method. We propose a new model named OvertoneNet (OveNet) that adopts two novel concepts, frequency 1x1 convolution layers, and complex-spectrogram channels, to handle the 44.1k audio signals (Hi-Res audio signals) containing full overtones. The result shows that OveNet performs well in both objective and subjective evaluation on interference using limited training data from SiSEC2018.",

keywords = "U-Net, audio source separation, convolutional neural networks",

author = "Wu, {Chi Sheng} and Shiang Lee and Soo, {Von Wun}",

note = "Publisher Copyright: {\textcopyright} 2019 IEEE.; 21st IEEE International Symposium on Multimedia, ISM 2019 ; Conference date: 09-12-2019 Through 11-12-2019",

year = "2019",

month = dec,

doi = "10.1109/ISM46123.2019.00034",

language = "英语",

series = "Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019",

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

pages = "148--151",

booktitle = "Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019",

address = "美国",

}

Wu, CS, Lee, S & Soo, VW 2019, OveNet: A Hyper-Range U-Net for Singing Voice Separation. in Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019., 8959050, Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019, Institute of Electrical and Electronics Engineers Inc., pp. 148-151, 21st IEEE International Symposium on Multimedia, ISM 2019, San Diego, United States, 09/12/19. https://doi.org/10.1109/ISM46123.2019.00034

OveNet: A Hyper-Range U-Net for Singing Voice Separation. / Wu, Chi Sheng; Lee, Shiang; Soo, Von Wun.
Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019. Institute of Electrical and Electronics Engineers Inc., 2019. p. 148-151 8959050 (Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019).

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

TY - GEN

T1 - OveNet

T2 - 21st IEEE International Symposium on Multimedia, ISM 2019

AU - Wu, Chi Sheng

AU - Lee, Shiang

AU - Soo, Von Wun

PY - 2019/12

Y1 - 2019/12

N2 - In the audio source separation topic, most researchers based on deep learning methods ignored higher-frequency signals due to lack of efficient data compression method. We propose a new model named OvertoneNet (OveNet) that adopts two novel concepts, frequency 1x1 convolution layers, and complex-spectrogram channels, to handle the 44.1k audio signals (Hi-Res audio signals) containing full overtones. The result shows that OveNet performs well in both objective and subjective evaluation on interference using limited training data from SiSEC2018.

AB - In the audio source separation topic, most researchers based on deep learning methods ignored higher-frequency signals due to lack of efficient data compression method. We propose a new model named OvertoneNet (OveNet) that adopts two novel concepts, frequency 1x1 convolution layers, and complex-spectrogram channels, to handle the 44.1k audio signals (Hi-Res audio signals) containing full overtones. The result shows that OveNet performs well in both objective and subjective evaluation on interference using limited training data from SiSEC2018.

KW - U-Net

KW - audio source separation

KW - convolutional neural networks

UR - http://www.scopus.com/inward/record.url?scp=85078888429&partnerID=8YFLogxK

U2 - 10.1109/ISM46123.2019.00034

DO - 10.1109/ISM46123.2019.00034

M3 - 会议稿件

AN - SCOPUS:85078888429

T3 - Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019

SP - 148

EP - 151

BT - Proceedings - 2019 IEEE International Symposium on Multimedia, ISM 2019

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 9 December 2019 through 11 December 2019

ER -

OveNet: A Hyper-Range U-Net for Singing Voice Separation

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this