Sound source separation by spectral subtraction based on instantaneous estimation of noise spectrum

Kenji Ozawa, Masanori Morise, Shuichi Sakamoto, Kanji Watanabe

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In our previous paper, we proposed a sound source separation method using the two-dimensional fast Fourier transform (2D FFT) of a spatio-temporal sound pressure distribution (STSPD) image that is composed from the outputs of a microphone array. In an STSPD image, vertical stripes are created for a target sound arriving from the perpendicular direction to the array; therefore, its spectral components are concentrated on the spatial direct current (DC) components in the 2D amplitude spectrum. In that study, we estimated the noise DC amplitudes using a deep neural network (DNN), then subtracted them from the observed spectrum to suppress the noise. However, the performance of noise suppression can be improved further. In this study, we estimate the noise DC components theoretically instead of empirically using a DNN. We improved the performance successfully.

Original languageEnglish
Title of host publication2019 6th International Conference on Systems and Informatics, ICSAI 2019
EditorsWanqing Wu, Lipo Wang, Chunlei Ji, Niansheng Chen, Sun Qiang, Xiaoyong Song, Xin Wang
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1137-1142
Number of pages6
ISBN (Electronic)9781728152561
DOIs
Publication statusPublished - 2019 Nov
Event6th International Conference on Systems and Informatics, ICSAI 2019 - Shanghai, China
Duration: 2019 Nov 22019 Nov 4

Publication series

Name2019 6th International Conference on Systems and Informatics, ICSAI 2019

Conference

Conference6th International Conference on Systems and Informatics, ICSAI 2019
CountryChina
CityShanghai
Period19/11/219/11/4

Keywords

  • Instantaneous estimation
  • Mathematical derivation
  • Spatio-temporal sound pressure distribution image
  • Spectral subtraction
  • Two-dimensional FFT

ASJC Scopus subject areas

  • Health Informatics
  • Computer Vision and Pattern Recognition
  • Information Systems
  • Control and Systems Engineering
  • Mechanical Engineering
  • Control and Optimization

Fingerprint Dive into the research topics of 'Sound source separation by spectral subtraction based on instantaneous estimation of noise spectrum'. Together they form a unique fingerprint.

  • Cite this

    Ozawa, K., Morise, M., Sakamoto, S., & Watanabe, K. (2019). Sound source separation by spectral subtraction based on instantaneous estimation of noise spectrum. In W. Wu, L. Wang, C. Ji, N. Chen, S. Qiang, X. Song, & X. Wang (Eds.), 2019 6th International Conference on Systems and Informatics, ICSAI 2019 (pp. 1137-1142). [9010477] (2019 6th International Conference on Systems and Informatics, ICSAI 2019). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICSAI48974.2019.9010477