A two-channel minimum mean-square error log-spectral amplitude estimator for speech enhancement

Min Seok Choi, Hong Goo Kang

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

10 Citations (Scopus)

Abstract

This paper proposes a novel two-channel speech enhancement structure using the minimum mean-square error log-spectral amplitude (MMSE-LSA) estimator. The proposed two-channel enhancement algorithm utilizes a spatial relationship between two input signals to accurately estimate the noise power spectral density (PSD) needed for the MMSE-LSA algorithm. The proposed structure improves the noise reduction capacity with less speech distortion, while its complexity is much lower than simple cascade structures. The performance of the proposed algorithm is evaluated by automatic speech recognition tests in a car environment. Comparing to a simple cascading of two- and single-channel algorithms, the proposed algorithm improves the relative recognition rate by 17.5 % for high speed conditions and 14.8 % for low speed conditions, respectively.

Original languageEnglish
Title of host publication2008 Hands-free Speech Communication and Microphone Arrays, Proceedings, HSCMA 2008
Pages152-155
Number of pages4
DOIs
Publication statusPublished - 2008 Sep 3
Event2008 Hands-free Speech Communication and Microphone Arrays, HSCMA 2008 - Trento, Italy
Duration: 2008 May 62008 May 8

Other

Other2008 Hands-free Speech Communication and Microphone Arrays, HSCMA 2008
CountryItaly
CityTrento
Period08/5/608/5/8

Fingerprint

Speech enhancement
Mean square error
Power spectral density
Noise abatement
Speech recognition
Railroad cars
performance

All Science Journal Classification (ASJC) codes

  • Hardware and Architecture
  • Electrical and Electronic Engineering
  • Communication

Cite this

Choi, M. S., & Kang, H. G. (2008). A two-channel minimum mean-square error log-spectral amplitude estimator for speech enhancement. In 2008 Hands-free Speech Communication and Microphone Arrays, Proceedings, HSCMA 2008 (pp. 152-155). [4538709] https://doi.org/10.1109/HSCMA.2008.4538709
Choi, Min Seok ; Kang, Hong Goo. / A two-channel minimum mean-square error log-spectral amplitude estimator for speech enhancement. 2008 Hands-free Speech Communication and Microphone Arrays, Proceedings, HSCMA 2008. 2008. pp. 152-155
@inproceedings{9dfb51dfb49d46ceb75503d8854662e3,
title = "A two-channel minimum mean-square error log-spectral amplitude estimator for speech enhancement",
abstract = "This paper proposes a novel two-channel speech enhancement structure using the minimum mean-square error log-spectral amplitude (MMSE-LSA) estimator. The proposed two-channel enhancement algorithm utilizes a spatial relationship between two input signals to accurately estimate the noise power spectral density (PSD) needed for the MMSE-LSA algorithm. The proposed structure improves the noise reduction capacity with less speech distortion, while its complexity is much lower than simple cascade structures. The performance of the proposed algorithm is evaluated by automatic speech recognition tests in a car environment. Comparing to a simple cascading of two- and single-channel algorithms, the proposed algorithm improves the relative recognition rate by 17.5 {\%} for high speed conditions and 14.8 {\%} for low speed conditions, respectively.",
author = "Choi, {Min Seok} and Kang, {Hong Goo}",
year = "2008",
month = "9",
day = "3",
doi = "10.1109/HSCMA.2008.4538709",
language = "English",
isbn = "9781424423385",
pages = "152--155",
booktitle = "2008 Hands-free Speech Communication and Microphone Arrays, Proceedings, HSCMA 2008",

}

Choi, MS & Kang, HG 2008, A two-channel minimum mean-square error log-spectral amplitude estimator for speech enhancement. in 2008 Hands-free Speech Communication and Microphone Arrays, Proceedings, HSCMA 2008., 4538709, pp. 152-155, 2008 Hands-free Speech Communication and Microphone Arrays, HSCMA 2008, Trento, Italy, 08/5/6. https://doi.org/10.1109/HSCMA.2008.4538709

A two-channel minimum mean-square error log-spectral amplitude estimator for speech enhancement. / Choi, Min Seok; Kang, Hong Goo.

2008 Hands-free Speech Communication and Microphone Arrays, Proceedings, HSCMA 2008. 2008. p. 152-155 4538709.

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

TY - GEN

T1 - A two-channel minimum mean-square error log-spectral amplitude estimator for speech enhancement

AU - Choi, Min Seok

AU - Kang, Hong Goo

PY - 2008/9/3

Y1 - 2008/9/3

N2 - This paper proposes a novel two-channel speech enhancement structure using the minimum mean-square error log-spectral amplitude (MMSE-LSA) estimator. The proposed two-channel enhancement algorithm utilizes a spatial relationship between two input signals to accurately estimate the noise power spectral density (PSD) needed for the MMSE-LSA algorithm. The proposed structure improves the noise reduction capacity with less speech distortion, while its complexity is much lower than simple cascade structures. The performance of the proposed algorithm is evaluated by automatic speech recognition tests in a car environment. Comparing to a simple cascading of two- and single-channel algorithms, the proposed algorithm improves the relative recognition rate by 17.5 % for high speed conditions and 14.8 % for low speed conditions, respectively.

AB - This paper proposes a novel two-channel speech enhancement structure using the minimum mean-square error log-spectral amplitude (MMSE-LSA) estimator. The proposed two-channel enhancement algorithm utilizes a spatial relationship between two input signals to accurately estimate the noise power spectral density (PSD) needed for the MMSE-LSA algorithm. The proposed structure improves the noise reduction capacity with less speech distortion, while its complexity is much lower than simple cascade structures. The performance of the proposed algorithm is evaluated by automatic speech recognition tests in a car environment. Comparing to a simple cascading of two- and single-channel algorithms, the proposed algorithm improves the relative recognition rate by 17.5 % for high speed conditions and 14.8 % for low speed conditions, respectively.

UR - http://www.scopus.com/inward/record.url?scp=50449084489&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=50449084489&partnerID=8YFLogxK

U2 - 10.1109/HSCMA.2008.4538709

DO - 10.1109/HSCMA.2008.4538709

M3 - Conference contribution

SN - 9781424423385

SP - 152

EP - 155

BT - 2008 Hands-free Speech Communication and Microphone Arrays, Proceedings, HSCMA 2008

ER -

Choi MS, Kang HG. A two-channel minimum mean-square error log-spectral amplitude estimator for speech enhancement. In 2008 Hands-free Speech Communication and Microphone Arrays, Proceedings, HSCMA 2008. 2008. p. 152-155. 4538709 https://doi.org/10.1109/HSCMA.2008.4538709