DMCA
Speaker verification using Adapted Gaussian mixture models (2000)
Cached
Download Links
| Venue: | Digital Signal Processing |
| Citations: | 1010 - 42 self |
BibTeX
@INPROCEEDINGS{Reynolds00speakerverification,
author = {Douglas A. Reynolds and Thomas F. Quatieri and Robert B. Dunn},
title = {Speaker verification using Adapted Gaussian mixture models},
booktitle = {Digital Signal Processing},
year = {2000},
pages = {2000}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
In this paper we describe the major elements of MIT Lincoln Laboratory’s Gaussian mixture model (GMM)-based speaker verification system used successfully in several NIST Speaker Recognition Evaluations (SREs). The system is built around the likelihood ratio test for verification, using simple but effective GMMs for likelihood functions, a universal background model (UBM) for alternative speaker representation, and a form of Bayesian adaptation to derive speaker models from the UBM. The development and use of a handset detector and score normalization to greatly improve verification performance is also described and discussed. Finally, representative performance benchmarks and system behavior experiments on NIST SRE corpora are presented. © 2000 Academic Press Key Words: speaker recognition; Gaussian mixture models; likelihood ratio detector; universal background model; handset normalization; NIST evaluation. 1.
Keyphrases
speaker verification adapted gaussian mixture model handset detector major element representative performance benchmark system behavior experiment verification performance bayesian adaptation likelihood function academic press key word universal background model likelihood ratio test speaker model nist evaluation speaker verification system speaker recognition several nist speaker recognition evaluation effective gmms gaussian mixture model alternative speaker representation nist sre corpus score normalization
