Improved wavelet based a-priori SNR estimation for speech enhancement

Pak Kong Lun; Tai Chiu Hsung

doi:10.1109/ISCAS.2010.5537182

Improved wavelet based a-priori SNR estimation for speech enhancement

Pak Kong Lun, Tai Chiu Hsung

The Hong Kong Polytechnic University

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

9 Citations (Scopus)

Abstract

To obtain a reliable estimate of the a-priori signal to noise (SNR) ratio is crucial to most frequency domain speech enhancement algorithms. Recently, the low variance multitaper spectrum (MTS) estimator with wavelet denoising was suggested for the estimation of the a-priori SNR. However, traditional approach directly plugs in the wavelet shrinkage denoiser and adopts the universal threshold which is not fully optimized to the characteristic of the MTS of noisy signals. In this paper, a twostage estimation algorithm is proposed. First, the log MTS components that are dominated by noise are detected and removed in the wavelet domain. Second, a modified SUREshrink scheme is applied to further remove the noise remained in the speech spectral peaks. The new estimator is applied to the traditional Wiener filter and log MMSE speech enhancement algorithms and leads to significantly better performance.

Original language	English
Title of host publication	ISCAS 2010 - 2010 IEEE International Symposium on Circuits and Systems
Subtitle of host publication	Nano-Bio Circuit Fabrics and Systems
Pages	2382-2385
Number of pages	4
DOIs	https://doi.org/10.1109/ISCAS.2010.5537182
Publication status	Published - 31 Aug 2010
Event	2010 IEEE International Symposium on Circuits and Systems: Nano-Bio Circuit Fabrics and Systems, ISCAS 2010 - Paris, France Duration: 30 May 2010 → 2 Jun 2010

Conference

Conference	2010 IEEE International Symposium on Circuits and Systems: Nano-Bio Circuit Fabrics and Systems, ISCAS 2010
Country/Territory	France
City	Paris
Period	30/05/10 → 2/06/10

ASJC Scopus subject areas

Hardware and Architecture
Electrical and Electronic Engineering

More information

10.1109/ISCAS.2010.5537182

Cite this

@inproceedings{28d6d0e85cd4483d889a4598c2a1f0d7,

title = "Improved wavelet based a-priori SNR estimation for speech enhancement",

abstract = "To obtain a reliable estimate of the a-priori signal to noise (SNR) ratio is crucial to most frequency domain speech enhancement algorithms. Recently, the low variance multitaper spectrum (MTS) estimator with wavelet denoising was suggested for the estimation of the a-priori SNR. However, traditional approach directly plugs in the wavelet shrinkage denoiser and adopts the universal threshold which is not fully optimized to the characteristic of the MTS of noisy signals. In this paper, a twostage estimation algorithm is proposed. First, the log MTS components that are dominated by noise are detected and removed in the wavelet domain. Second, a modified SUREshrink scheme is applied to further remove the noise remained in the speech spectral peaks. The new estimator is applied to the traditional Wiener filter and log MMSE speech enhancement algorithms and leads to significantly better performance.",

author = "Lun, {Pak Kong} and Hsung, {Tai Chiu}",

year = "2010",

month = aug,

day = "31",

doi = "10.1109/ISCAS.2010.5537182",

language = "English",

isbn = "9781424453085",

pages = "2382--2385",

booktitle = "ISCAS 2010 - 2010 IEEE International Symposium on Circuits and Systems",

note = "2010 IEEE International Symposium on Circuits and Systems: Nano-Bio Circuit Fabrics and Systems, ISCAS 2010 ; Conference date: 30-05-2010 Through 02-06-2010",

}

Lun, PK & Hsung, TC 2010, Improved wavelet based a-priori SNR estimation for speech enhancement. in ISCAS 2010 - 2010 IEEE International Symposium on Circuits and Systems: Nano-Bio Circuit Fabrics and Systems., 5537182, pp. 2382-2385, 2010 IEEE International Symposium on Circuits and Systems: Nano-Bio Circuit Fabrics and Systems, ISCAS 2010, Paris, France, 30/05/10. https://doi.org/10.1109/ISCAS.2010.5537182

Improved wavelet based a-priori SNR estimation for speech enhancement. / Lun, Pak Kong; Hsung, Tai Chiu.
ISCAS 2010 - 2010 IEEE International Symposium on Circuits and Systems: Nano-Bio Circuit Fabrics and Systems. 2010. p. 2382-2385 5537182.

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

TY - GEN

T1 - Improved wavelet based a-priori SNR estimation for speech enhancement

AU - Lun, Pak Kong

AU - Hsung, Tai Chiu

PY - 2010/8/31

Y1 - 2010/8/31

N2 - To obtain a reliable estimate of the a-priori signal to noise (SNR) ratio is crucial to most frequency domain speech enhancement algorithms. Recently, the low variance multitaper spectrum (MTS) estimator with wavelet denoising was suggested for the estimation of the a-priori SNR. However, traditional approach directly plugs in the wavelet shrinkage denoiser and adopts the universal threshold which is not fully optimized to the characteristic of the MTS of noisy signals. In this paper, a twostage estimation algorithm is proposed. First, the log MTS components that are dominated by noise are detected and removed in the wavelet domain. Second, a modified SUREshrink scheme is applied to further remove the noise remained in the speech spectral peaks. The new estimator is applied to the traditional Wiener filter and log MMSE speech enhancement algorithms and leads to significantly better performance.

AB - To obtain a reliable estimate of the a-priori signal to noise (SNR) ratio is crucial to most frequency domain speech enhancement algorithms. Recently, the low variance multitaper spectrum (MTS) estimator with wavelet denoising was suggested for the estimation of the a-priori SNR. However, traditional approach directly plugs in the wavelet shrinkage denoiser and adopts the universal threshold which is not fully optimized to the characteristic of the MTS of noisy signals. In this paper, a twostage estimation algorithm is proposed. First, the log MTS components that are dominated by noise are detected and removed in the wavelet domain. Second, a modified SUREshrink scheme is applied to further remove the noise remained in the speech spectral peaks. The new estimator is applied to the traditional Wiener filter and log MMSE speech enhancement algorithms and leads to significantly better performance.

UR - http://www.scopus.com/inward/record.url?scp=77956002415&partnerID=8YFLogxK

U2 - 10.1109/ISCAS.2010.5537182

DO - 10.1109/ISCAS.2010.5537182

M3 - Conference article published in proceeding or book

SN - 9781424453085

SP - 2382

EP - 2385

BT - ISCAS 2010 - 2010 IEEE International Symposium on Circuits and Systems

T2 - 2010 IEEE International Symposium on Circuits and Systems: Nano-Bio Circuit Fabrics and Systems, ISCAS 2010

Y2 - 30 May 2010 through 2 June 2010

ER -

Improved wavelet based a-priori SNR estimation for speech enhancement

Abstract

Conference

ASJC Scopus subject areas

More information

Other files and links

Fingerprint

Cite this