Conference paper

 
Kockmann, M., Burget, L., Glembek, O., Ferrer, L., Černocký, J.: Prosodic Speaker Verification using Subspace Multinomial Models with Intersession Compensation, In: Proceedings of the 11th Annual Conference of the International Speech Communication Association (INTERSPEECH 2010), Makuhari, Chiba, Japan, JP, ISCA, 2010, p. 1061-1064, ISBN 978-1-61782-123-3, ISSN 1990-9772
Publication language:english
Original title:Prosodic Speaker Verification using Subspace Multinomial Models with Intersession Compensation
Title (cs):Prosodické ověřování mluvčího pomocí multinomiálních modelů v podprostorech s kompensací variability mezi nahrávkami
Pages:1061-1064
Proceedings:Proceedings of the 11th Annual Conference of the International Speech Communication Association (INTERSPEECH 2010)
Conference:Interspeech 2010
Place:Makuhari, Chiba, Japan, JP
Year:2010
ISBN:978-1-61782-123-3
Journal:Proceedings of Interspeech, Vol. 2010, No. 9, FR
ISSN:1990-9772
Publisher:International Speech Communication Association
URL:http://www.fit.vutbr.cz/research/groups/speech/publi/2010/kockmann_interspeech2010_IS100048.pdf [PDF]
Keywords
speaker verification, prosody, JFA, multinomial model
Annotation
The paper is on the proposal of a novel approach to modeling prosodic features. Our model is based on the idea of introducing subspace of model parameters.
Abstract
We propose a novel approach to modeling prosodic features. Inspired by Joint Factor Analysis model (JFA), our model is based on the same idea of introducing subspace of model parameters. However, the underlying Gaussian Mixture distribution of JFA is replaced by multinomial distribution to model sequences of discrete units rather than continuous features. In this work, we use the subspace model as a feature extractor for support vector machines (SVMs), similar to the recently proposed JFA in total variability space. We can show the capability to reduce high-dimensional count vectors to low dimension while keeping system performance stable. With additional intersession compensation, we can improve 30% relative to the baseline system and reach an equal error rate of 8.8% on the NIST 2006 SRE dataset.
BibTeX:
@INPROCEEDINGS{
   author = {Marcel Kockmann and Lukáš Burget and Ondřej Glembek and
	Luciana Ferrer and Jan Černocký},
   title = {Prosodic Speaker Verification using Subspace Multinomial
	Models with Intersession Compensation},
   pages = {1061--1064},
   booktitle = {Proceedings of the 11th Annual Conference of the
	International Speech Communication Association (INTERSPEECH
	2010)},
   journal = {Proceedings of Interspeech},
   volume = {2010},
   number = {9},
   year = {2010},
   location = {Makuhari, Chiba, Japan, JP},
   publisher = {International Speech Communication Association},
   ISBN = {978-1-61782-123-3},
   ISSN = {1990-9772},
   language = {english},
   url = {http://www.fit.vutbr.cz/research/view_pub.php?id=9360}
}