Noise variance estimation for kaiman filtering of noisy speech

Wooil Kim, Hanseok Ko

Research output: Contribution to journalArticle

24 Citations (Scopus)

Abstract

This paper proposes an algorithm that adaptively estimates time-varying noise variance used in Kaiman filtering for real-time speech signal enhancement. In the speech signal contaminated by white noise, the spectral components except dominant ones in high frequency band are expected to reflect the noise energy. Our approach is first to find the dominant energy bands over speech spectrum using LPC. We then calculate the average value of the actual spectral components over the high frequency region excluding the dominant energy bands and use it as the noise variance. The resulting noise variance estimate is then applied to Kaiman filtering to suppress the background noise. Experimental results indicate that the proposed approach achieves a significant improvement in terms of speech enhancement over those of the conventional Kaiman filtering that uses the average noise power over silence interval only. As a refinement of our results, we employ multiple-Kalman filtering with multiple noise models and improve the intelligibility.

Original languageEnglish
Pages (from-to)155-160
Number of pages6
JournalIEICE Transactions on Information and Systems
VolumeE84-D
Issue number1
Publication statusPublished - 2001 Dec 1

Fingerprint

Band structure
Speech enhancement
White noise
Frequency bands

Keywords

  • Dominant energy band
  • Enhancement
  • Kaiman filtering
  • Multiple kaiman filtering
  • Noise variance
  • Speech

ASJC Scopus subject areas

  • Computer Graphics and Computer-Aided Design
  • Information Systems
  • Software

Cite this

Noise variance estimation for kaiman filtering of noisy speech. / Kim, Wooil; Ko, Hanseok.

In: IEICE Transactions on Information and Systems, Vol. E84-D, No. 1, 01.12.2001, p. 155-160.

Research output: Contribution to journalArticle

@article{a3c162d4da4a4e35b906772c92577585,
title = "Noise variance estimation for kaiman filtering of noisy speech",
abstract = "This paper proposes an algorithm that adaptively estimates time-varying noise variance used in Kaiman filtering for real-time speech signal enhancement. In the speech signal contaminated by white noise, the spectral components except dominant ones in high frequency band are expected to reflect the noise energy. Our approach is first to find the dominant energy bands over speech spectrum using LPC. We then calculate the average value of the actual spectral components over the high frequency region excluding the dominant energy bands and use it as the noise variance. The resulting noise variance estimate is then applied to Kaiman filtering to suppress the background noise. Experimental results indicate that the proposed approach achieves a significant improvement in terms of speech enhancement over those of the conventional Kaiman filtering that uses the average noise power over silence interval only. As a refinement of our results, we employ multiple-Kalman filtering with multiple noise models and improve the intelligibility.",
keywords = "Dominant energy band, Enhancement, Kaiman filtering, Multiple kaiman filtering, Noise variance, Speech",
author = "Wooil Kim and Hanseok Ko",
year = "2001",
month = "12",
day = "1",
language = "English",
volume = "E84-D",
pages = "155--160",
journal = "IEICE Transactions on Information and Systems",
issn = "0916-8532",
publisher = "Maruzen Co., Ltd/Maruzen Kabushikikaisha",
number = "1",

}

TY - JOUR

T1 - Noise variance estimation for kaiman filtering of noisy speech

AU - Kim, Wooil

AU - Ko, Hanseok

PY - 2001/12/1

Y1 - 2001/12/1

N2 - This paper proposes an algorithm that adaptively estimates time-varying noise variance used in Kaiman filtering for real-time speech signal enhancement. In the speech signal contaminated by white noise, the spectral components except dominant ones in high frequency band are expected to reflect the noise energy. Our approach is first to find the dominant energy bands over speech spectrum using LPC. We then calculate the average value of the actual spectral components over the high frequency region excluding the dominant energy bands and use it as the noise variance. The resulting noise variance estimate is then applied to Kaiman filtering to suppress the background noise. Experimental results indicate that the proposed approach achieves a significant improvement in terms of speech enhancement over those of the conventional Kaiman filtering that uses the average noise power over silence interval only. As a refinement of our results, we employ multiple-Kalman filtering with multiple noise models and improve the intelligibility.

AB - This paper proposes an algorithm that adaptively estimates time-varying noise variance used in Kaiman filtering for real-time speech signal enhancement. In the speech signal contaminated by white noise, the spectral components except dominant ones in high frequency band are expected to reflect the noise energy. Our approach is first to find the dominant energy bands over speech spectrum using LPC. We then calculate the average value of the actual spectral components over the high frequency region excluding the dominant energy bands and use it as the noise variance. The resulting noise variance estimate is then applied to Kaiman filtering to suppress the background noise. Experimental results indicate that the proposed approach achieves a significant improvement in terms of speech enhancement over those of the conventional Kaiman filtering that uses the average noise power over silence interval only. As a refinement of our results, we employ multiple-Kalman filtering with multiple noise models and improve the intelligibility.

KW - Dominant energy band

KW - Enhancement

KW - Kaiman filtering

KW - Multiple kaiman filtering

KW - Noise variance

KW - Speech

UR - http://www.scopus.com/inward/record.url?scp=0035122884&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0035122884&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:0035122884

VL - E84-D

SP - 155

EP - 160

JO - IEICE Transactions on Information and Systems

JF - IEICE Transactions on Information and Systems

SN - 0916-8532

IS - 1

ER -