Faculty of Information Technology, BUT

Publication Details

Combination of Speech Features Using Smoothed Heteroscedastic Linear Discriminant Analysis

BURGET Lukáš. Combination of Speech Features Using Smoothed Heteroscedastic Linear Discriminant Analysis. In: Proc. 8th International Conference on Spoken Language Processing. Jeju island: Sunjin Printing Co,, 2004, pp. 2549-2552.
Czech title
Kombinování řečových příznaků použitím vyhlazené heteroscedastické diskriminační analýzy
Type
conference paper
Language
english
Authors
URL
Keywords
speech recognition, LDA, HLDA, feature extraction, feature combination
Abstract
Combination of Speech Features Using Smoothed Heteroscedastic Linear Discriminant Analysis
Annotation
Feature combination techniques based on PCA, LDA and HLDA are compared in experiments where limited amount of training data is available. Success with feature combination can be quite dependent on proper estimation of statistics required by the used technique. Insufficiency of training data is, therefore, an important problem, which has to be taken in to account in our experiments.
Besides of some standard approaches increasing robustness of statistic estimation, methods based on combination of LDA and HLDA are proposed. An improved recognition performance obtained using these methods is demonstrated in experiments.
Published
2004
Pages
2549-2552
Proceedings
Proc. 8th International Conference on Spoken Language Processing
Conference
8th International Conference on Spoken Language Processing, Jeju Island, KR
Publisher
Sunjin Printing Co,
Place
Jeju island, KR
BibTeX
@INPROCEEDINGS{FITPUB7486,
   author = "Luk\'{a}\v{s} Burget",
   title = "Combination of Speech Features Using Smoothed Heteroscedastic Linear Discriminant Analysis",
   pages = "2549--2552",
   booktitle = "Proc. 8th International Conference on Spoken Language Processing",
   year = 2004,
   location = "Jeju island, KR",
   publisher = "Sunjin Printing Co,",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/7486"
}
Back to top