Comparison of generative and discriminative approaches for speaker recognition with limited data

Simple item page
dc.contributor.authorSilovský, Jan
dc.contributor.authorČerva, Petr
dc.contributor.authorŽďánský, Jindřich
dc.date.accessioned2016-07-08
dc.date.available2016-07-08
dc.date.issued2009
dc.description.abstractThis paper presents a comparison of three different speaker recognition methods deployed in a broadcast news processing system. We focus on how the generative and discriminative nature of these methods affects the speaker recognition framework and we also deal with intersession variability compensation techniques in more detail, which are of great interest in broadcast processing domain. Performed experiments are specific particularly for the very limited amount of data used for both speaker enrollment (typically ranging from 30 to 60 seconds) and recognition (typically ranging from 5 to 15 seconds). Our results show that the system based on Gaussian Mixture Models (GMMs) outperforms both systems based on Support Vector Machines (SVMs) but its drawback is higher computational cost.en
dc.formattext
dc.identifier.issn1210-2512
dc.identifier.scopus2-s2.0-77951188220
dc.identifier.urihttps://dspace.tul.cz/handle/15240/16641
dc.language.isoen
dc.relation.ispartofRadioengineering
dc.sourcej-scopus
dc.sourcej-wok
dc.subjectBroadcast processingen
dc.subjectGaussian Mixture Models (GMM)en
dc.subjectSpeaker recognitionen
dc.subjectSupport Vector Machines (SVM)en
dc.titleComparison of generative and discriminative approaches for speaker recognition with limited dataen
dc.typearticle
local.accessopen access
local.citation.epage316
local.citation.spage307
local.departmentSpeechLab
local.facultyFaculty of Mechatronic and Interdisciplinary Studies
local.fulltextyes
local.identifier.wok269744600008
local.notenefunguje RIV
local.relation.issue2-1
local.relation.volume18
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
09_03_307_316.pdf
Size:
3.56 MB
Format:
Adobe Portable Document Format
Description:
Download
Collections