Skip to content

Publications

Detail of publication

Citation

Jan Vaněk and Lukáš Machlica and Josef V. Psutka and Josef Psutka : Covariance Matrix Enhancement Approach to Train Robust Gaussian Mixture Models of Speech Data . Speech and Computer, Lecture Notes in Computer Science, vol. 8113, p. 92-99, Springer, 2013.

Download PDF

PDF

Abstract

An estimation of parameters of a multivariate Gaussian Mixture Model is usually based on a criterion (e.g. Maximum Likelihood) that is focused mostly on training data. Therefore, testing data, which were not seen during the training procedure, may cause problems. Moreover, numerical instabilities can occur (e.g. for low-occupied Gaussians especially when working with full-covariance matrices in high-dimensional spaces). Another question concerns the number of Gaussians to be trained for a specific data set. The approach proposed in this paper can handle all these issues. It is based on an assumption that the training and testing data were generated from the same source distribution. The key part of the approach is to use a criterion based on the source distribution rather than using the training data itself. It is shown how to modify an estimation procedure in order to fit the source distribution better (despite the fact that it is unknown), and subsequently new estimation algorithm for diagonal- as well as full-covariance matrices is derived and tested.

Detail of publication

Title: Covariance Matrix Enhancement Approach to Train Robust Gaussian Mixture Models of Speech Data
Author: Jan Vaněk ; Lukáš Machlica ; Josef V. Psutka ; Josef Psutka
Language: English
Date of publication: 1 Sep 2013
Year: 2013
Type of publication: Papers in proceedings of reviewed conferences
Book title: Speech and Computer
Series: Lecture Notes in Computer Science
Číslo vydání: 8113
Page: 92 - 99
DOI: 10.1007/978-3-319-01931-4_13
ISBN: 978-3-319-01930-7
Publisher: Springer
/ 2014-11-12 12:21:45 /

Keywords

Gaussian Mixture Models, Full Covariance, Full Covariance Matrix, Regularization, Automatic Speech Recognition

BibTeX

@INPROCEEDINGS{JanVanek_2013_CovarianceMatrix,
 author = {Jan Van\v{e}k and Luk\'{a}\v{s} Machlica and Josef V. Psutka and Josef Psutka},
 title = {Covariance Matrix Enhancement Approach to Train Robust Gaussian Mixture Models of Speech Data},
 year = {2013},
 publisher = {Springer},
 volume = {8113},
 pages = {92-99},
 booktitle = {Speech and Computer},
 series = {Lecture Notes in Computer Science},
 ISBN = {978-3-319-01930-7},
 doi = {10.1007/978-3-319-01931-4_13},
 url = {http://www.kky.zcu.cz/en/publications/JanVanek_2013_CovarianceMatrix},
}