Analysis of Feature Extraction and Channel Compensation in a GMM Speaker Recognition System

Authors:
L. Burget;P. Matejka;P. Schwarz;O. Glembek;J. Cernocky
Affiliations:
Brno Univ. of Technol., Brno;-;-;-;-
Venue:
IEEE Transactions on Audio, Speech, and Language Processing
Year:
2007

Citing 0
Cited 9

Ubiquitous and Robust Text-Independent Speaker Recognition for Home Automation Digital Life

UIC '08 Proceedings of the 5th international conference on Ubiquitous Intelligence and Computing
Acquisition of Telephone Data from Radio Broadcasts with Applications to Language Recognition

TSD '08 Proceedings of the 11th international conference on Text, Speech and Dialogue
The likelihood ratio decision criterion for nuisance attribute projection in GMM speaker verification

EURASIP Journal on Advances in Signal Processing
Particle swarm optimization for sorted adapted Gaussian mixture models

IEEE Transactions on Audio, Speech, and Language Processing
Modulation spectral features for robust far-field speaker identification

IEEE Transactions on Audio, Speech, and Language Processing
Multimodal speaker verification based on electroglottograph signal and glottal activity detection

EURASIP Journal on Advances in Signal Processing
Comparison between supervised and unsupervised learning of probabilistic linear discriminant analysis mixture models for speaker verification

Pattern Recognition Letters
Intoxicated speech detection: A fusion framework with speaker-normalized hierarchical functionals and GMM supervectors

Computer Speech and Language
Granular Computing Based on Gaussian Cloud Transformation

Fundamenta Informaticae - To Andrzej Skowron on His 70th Birthday

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, several feature extraction and channel compensation techniques found in state-of-the-art speaker verification systems are analyzed and discussed. For the NIST SRE 2006 submission, cepstral mean subtraction, feature warping, RelAtive SpecTrAl (RASTA) filtering, heteroscedastic linear discriminant analysis (HLDA), feature mapping, and eigenchannel adaptation were incrementally added to minimize the system's error rate. This paper deals with eigenchannel adaptation in more detail and includes its theoretical background and implementation issues. The key part of the paper is, however, the post-evaluation analysis, undermining a common myth that ldquothe more boxes in the scheme, the better the system.rdquo All results are presented on NIST Speaker Recognition Evaluation (SRE) 2005 and 2006 data.