Robust model adaptation using mean and variance transformations in linear spectral domain

Donghyun Kim, Dongsuk Yook

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

1 Citation (Scopus)

Abstract

In this paper, we propose robust speech adaptation technique using continuous density hidden Markov models (HMMs) in unknown environments. This adaptation technique is an improved maximum likelihood linear spectral transformation (ML-LST) method, which aims to find appropriate noise parameters in the linear spectral domain. Previously, ML-LST and many transform-based adaptation algorithms have been applied to the Gaussian mean vectors of HMM systems. In the improved ML-LST for the rapid adaptation, the mean vectors and covariance matrices of an HMM based speech recognizer are transformed simultaneously using a small number of transformation parameters. It is shown that the variance transformation provides important information which can be used to handle environmental noise, in the similar manner that the mean transformation does.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science
EditorsM. Gallagher, J. Hogan, F. Maire
Pages149-154
Number of pages6
Volume3578
Publication statusPublished - 2005
Event6th International Conference on Intelligent Data Engineering and Automated Learning - IDEAL 2005 - Brisbane, Australia
Duration: 2005 Jul 62005 Jul 8

Other

Other6th International Conference on Intelligent Data Engineering and Automated Learning - IDEAL 2005
CountryAustralia
CityBrisbane
Period05/7/605/7/8

Fingerprint

Hidden Markov models
Maximum likelihood
Covariance matrix
Mathematical transformations

ASJC Scopus subject areas

  • Computer Science (miscellaneous)

Cite this

Kim, D., & Yook, D. (2005). Robust model adaptation using mean and variance transformations in linear spectral domain. In M. Gallagher, J. Hogan, & F. Maire (Eds.), Lecture Notes in Computer Science (Vol. 3578, pp. 149-154)

Robust model adaptation using mean and variance transformations in linear spectral domain. / Kim, Donghyun; Yook, Dongsuk.

Lecture Notes in Computer Science. ed. / M. Gallagher; J. Hogan; F. Maire. Vol. 3578 2005. p. 149-154.

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

Kim, D & Yook, D 2005, Robust model adaptation using mean and variance transformations in linear spectral domain. in M Gallagher, J Hogan & F Maire (eds), Lecture Notes in Computer Science. vol. 3578, pp. 149-154, 6th International Conference on Intelligent Data Engineering and Automated Learning - IDEAL 2005, Brisbane, Australia, 05/7/6.
Kim D, Yook D. Robust model adaptation using mean and variance transformations in linear spectral domain. In Gallagher M, Hogan J, Maire F, editors, Lecture Notes in Computer Science. Vol. 3578. 2005. p. 149-154
Kim, Donghyun ; Yook, Dongsuk. / Robust model adaptation using mean and variance transformations in linear spectral domain. Lecture Notes in Computer Science. editor / M. Gallagher ; J. Hogan ; F. Maire. Vol. 3578 2005. pp. 149-154
@inproceedings{50ce4d39b3994333b85447efd4e1b9eb,
title = "Robust model adaptation using mean and variance transformations in linear spectral domain",
abstract = "In this paper, we propose robust speech adaptation technique using continuous density hidden Markov models (HMMs) in unknown environments. This adaptation technique is an improved maximum likelihood linear spectral transformation (ML-LST) method, which aims to find appropriate noise parameters in the linear spectral domain. Previously, ML-LST and many transform-based adaptation algorithms have been applied to the Gaussian mean vectors of HMM systems. In the improved ML-LST for the rapid adaptation, the mean vectors and covariance matrices of an HMM based speech recognizer are transformed simultaneously using a small number of transformation parameters. It is shown that the variance transformation provides important information which can be used to handle environmental noise, in the similar manner that the mean transformation does.",
author = "Donghyun Kim and Dongsuk Yook",
year = "2005",
language = "English",
volume = "3578",
pages = "149--154",
editor = "M. Gallagher and J. Hogan and F. Maire",
booktitle = "Lecture Notes in Computer Science",

}

TY - GEN

T1 - Robust model adaptation using mean and variance transformations in linear spectral domain

AU - Kim, Donghyun

AU - Yook, Dongsuk

PY - 2005

Y1 - 2005

N2 - In this paper, we propose robust speech adaptation technique using continuous density hidden Markov models (HMMs) in unknown environments. This adaptation technique is an improved maximum likelihood linear spectral transformation (ML-LST) method, which aims to find appropriate noise parameters in the linear spectral domain. Previously, ML-LST and many transform-based adaptation algorithms have been applied to the Gaussian mean vectors of HMM systems. In the improved ML-LST for the rapid adaptation, the mean vectors and covariance matrices of an HMM based speech recognizer are transformed simultaneously using a small number of transformation parameters. It is shown that the variance transformation provides important information which can be used to handle environmental noise, in the similar manner that the mean transformation does.

AB - In this paper, we propose robust speech adaptation technique using continuous density hidden Markov models (HMMs) in unknown environments. This adaptation technique is an improved maximum likelihood linear spectral transformation (ML-LST) method, which aims to find appropriate noise parameters in the linear spectral domain. Previously, ML-LST and many transform-based adaptation algorithms have been applied to the Gaussian mean vectors of HMM systems. In the improved ML-LST for the rapid adaptation, the mean vectors and covariance matrices of an HMM based speech recognizer are transformed simultaneously using a small number of transformation parameters. It is shown that the variance transformation provides important information which can be used to handle environmental noise, in the similar manner that the mean transformation does.

UR - http://www.scopus.com/inward/record.url?scp=26444442342&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=26444442342&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:26444442342

VL - 3578

SP - 149

EP - 154

BT - Lecture Notes in Computer Science

A2 - Gallagher, M.

A2 - Hogan, J.

A2 - Maire, F.

ER -