Publication Details

Analysis of feature extraction and channel compensation in GMM speaker recognition system

BURGET Lukáš, MATĚJKA Pavel, SCHWARZ Petr, GLEMBEK Ondřej and ČERNOCKÝ Jan. Analysis of feature extraction and channel compensation in GMM speaker recognition system. IEEE Transactions on Audio, Speech, and Language Processing, vol. 15, no. 7, 2007, pp. 1979-1986. ISSN 1558-7916.
Czech title
Analýza výpočtu příznaků a kompenzace na kanál v GMM systému pro rozpoznávání mluvčího
Type
journal article
Language
english
Authors
Burget Lukáš, doc. Ing., Ph.D. (DCGM FIT BUT)
Matějka Pavel, Ing., Ph.D. (DCGM FIT BUT)
Schwarz Petr, Ing., Ph.D. (DCGM FIT BUT)
Glembek Ondřej, Ing., Ph.D. (DCGM FIT BUT)
Černocký Jan, prof. Dr. Ing. (DCGM FIT BUT)
URL
Keywords

Speaker recognition, GMM, feature warping, RASTA, HLDA, Feature Mapping, eigenchannel adaptation.

Abstract

The paper is about Analysis of feature extraction and channel compensation in GMM speaker recognition system.

Annotation

In this paper, several feature extraction and channel compensation techniques found in state-of-the-art speaker verification systems are analyzed and discussed. For NIST SRE 2006 submission, Cepstral Mean Subtraction, Feature Warping, RASTA filtering, HLDA, Feature Mapping and eigenchannel adaptation were incrementally added to minimize system's error rate. The key-part of the paper is however the post-evaluation analysis, undermining the common myth "the more boxes in the scheme, the better system". All results are presented on NIST SRE 2005 and 2006 data.

Published
2007
Pages
1979-1986
Journal
IEEE Transactions on Audio, Speech, and Language Processing, vol. 15, no. 7, ISSN 1558-7916
Publisher
IEEE Signal Processing Society
BibTeX 
@ARTICLE{FITPUB8267,
   author = "Luk\'{a}\v{s} Burget and Pavel Mat\v{e}jka and Petr Schwarz and Ond\v{r}ej Glembek and Jan \v{C}ernock\'{y}",
   title = "Analysis of feature extraction and channel compensation in GMM speaker recognition system",
   pages = "1979--1986",
   journal = "IEEE Transactions on Audio, Speech, and Language Processing",
   volume = 15,
   number = 7,
   year = 2007,
   ISSN = "1558-7916",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/8267"
}
Back to top