Adaptation of HMM dynamic parameters in reverberant environment

Jinkyu Lee, Hyunson Seo, Hong Goo Kang

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

Abstract

This paper presents a new adaptation method for HMM-based automatic speech recognition system in a reverberant environment. Unlike the conventional approach that estimates dynamic mean vectors by adopting a spline interpolation technique, the proposed algorithm uses the transform derived by the mathematical property. Additionally, we introduce the adaptation for covariance matrices with the domain conversion process induced by log-normal distribution, because the statistical parameters are affected by not only mean vectors but also covariance matrices. Consequently, all statistical parameters in HMM can be adapted by the exact same transform structure. Experimental results show that the proposed method improves the recognition rate, in spite of having much simple adaptation process. Also it is robust to the estimation error that is unavoidable while extracting the reverberation time related parameters.

Original languageEnglish
Title of host publication2013 Proceedings of the 21st European Signal Processing Conference, EUSIPCO 2013
PublisherEuropean Signal Processing Conference, EUSIPCO
ISBN (Print)9780992862602
Publication statusPublished - 2013 Jan 1
Event2013 21st European Signal Processing Conference, EUSIPCO 2013 - Marrakech, Morocco
Duration: 2013 Sep 92013 Sep 13

Publication series

NameEuropean Signal Processing Conference
ISSN (Print)2219-5491

Other

Other2013 21st European Signal Processing Conference, EUSIPCO 2013
CountryMorocco
CityMarrakech
Period13/9/913/9/13

Fingerprint

Covariance matrix
Reverberation
Normal distribution
Speech recognition
Splines
Error analysis
Interpolation

All Science Journal Classification (ASJC) codes

  • Signal Processing
  • Electrical and Electronic Engineering

Cite this

Lee, J., Seo, H., & Kang, H. G. (2013). Adaptation of HMM dynamic parameters in reverberant environment. In 2013 Proceedings of the 21st European Signal Processing Conference, EUSIPCO 2013 [6811763] (European Signal Processing Conference). European Signal Processing Conference, EUSIPCO.
Lee, Jinkyu ; Seo, Hyunson ; Kang, Hong Goo. / Adaptation of HMM dynamic parameters in reverberant environment. 2013 Proceedings of the 21st European Signal Processing Conference, EUSIPCO 2013. European Signal Processing Conference, EUSIPCO, 2013. (European Signal Processing Conference).
@inproceedings{138b4c5d5da94b03947c8c6f2315f14d,
title = "Adaptation of HMM dynamic parameters in reverberant environment",
abstract = "This paper presents a new adaptation method for HMM-based automatic speech recognition system in a reverberant environment. Unlike the conventional approach that estimates dynamic mean vectors by adopting a spline interpolation technique, the proposed algorithm uses the transform derived by the mathematical property. Additionally, we introduce the adaptation for covariance matrices with the domain conversion process induced by log-normal distribution, because the statistical parameters are affected by not only mean vectors but also covariance matrices. Consequently, all statistical parameters in HMM can be adapted by the exact same transform structure. Experimental results show that the proposed method improves the recognition rate, in spite of having much simple adaptation process. Also it is robust to the estimation error that is unavoidable while extracting the reverberation time related parameters.",
author = "Jinkyu Lee and Hyunson Seo and Kang, {Hong Goo}",
year = "2013",
month = "1",
day = "1",
language = "English",
isbn = "9780992862602",
series = "European Signal Processing Conference",
publisher = "European Signal Processing Conference, EUSIPCO",
booktitle = "2013 Proceedings of the 21st European Signal Processing Conference, EUSIPCO 2013",

}

Lee, J, Seo, H & Kang, HG 2013, Adaptation of HMM dynamic parameters in reverberant environment. in 2013 Proceedings of the 21st European Signal Processing Conference, EUSIPCO 2013., 6811763, European Signal Processing Conference, European Signal Processing Conference, EUSIPCO, 2013 21st European Signal Processing Conference, EUSIPCO 2013, Marrakech, Morocco, 13/9/9.

Adaptation of HMM dynamic parameters in reverberant environment. / Lee, Jinkyu; Seo, Hyunson; Kang, Hong Goo.

2013 Proceedings of the 21st European Signal Processing Conference, EUSIPCO 2013. European Signal Processing Conference, EUSIPCO, 2013. 6811763 (European Signal Processing Conference).

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

TY - GEN

T1 - Adaptation of HMM dynamic parameters in reverberant environment

AU - Lee, Jinkyu

AU - Seo, Hyunson

AU - Kang, Hong Goo

PY - 2013/1/1

Y1 - 2013/1/1

N2 - This paper presents a new adaptation method for HMM-based automatic speech recognition system in a reverberant environment. Unlike the conventional approach that estimates dynamic mean vectors by adopting a spline interpolation technique, the proposed algorithm uses the transform derived by the mathematical property. Additionally, we introduce the adaptation for covariance matrices with the domain conversion process induced by log-normal distribution, because the statistical parameters are affected by not only mean vectors but also covariance matrices. Consequently, all statistical parameters in HMM can be adapted by the exact same transform structure. Experimental results show that the proposed method improves the recognition rate, in spite of having much simple adaptation process. Also it is robust to the estimation error that is unavoidable while extracting the reverberation time related parameters.

AB - This paper presents a new adaptation method for HMM-based automatic speech recognition system in a reverberant environment. Unlike the conventional approach that estimates dynamic mean vectors by adopting a spline interpolation technique, the proposed algorithm uses the transform derived by the mathematical property. Additionally, we introduce the adaptation for covariance matrices with the domain conversion process induced by log-normal distribution, because the statistical parameters are affected by not only mean vectors but also covariance matrices. Consequently, all statistical parameters in HMM can be adapted by the exact same transform structure. Experimental results show that the proposed method improves the recognition rate, in spite of having much simple adaptation process. Also it is robust to the estimation error that is unavoidable while extracting the reverberation time related parameters.

UR - http://www.scopus.com/inward/record.url?scp=84901344219&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84901344219&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84901344219

SN - 9780992862602

T3 - European Signal Processing Conference

BT - 2013 Proceedings of the 21st European Signal Processing Conference, EUSIPCO 2013

PB - European Signal Processing Conference, EUSIPCO

ER -

Lee J, Seo H, Kang HG. Adaptation of HMM dynamic parameters in reverberant environment. In 2013 Proceedings of the 21st European Signal Processing Conference, EUSIPCO 2013. European Signal Processing Conference, EUSIPCO. 2013. 6811763. (European Signal Processing Conference).