Full metadata record
DC FieldValueLanguage
dc.contributor.authorZajíc, Zbyněk
dc.contributor.authorMachlica, Lukáš
dc.contributor.authorMüller, Luděk
dc.date.accessioned2015-12-10T09:58:30Z-
dc.date.available2015-12-10T09:58:30Z-
dc.date.issued2011
dc.identifier.isbn978-3-642-23537-5
dc.identifier.issn0302-9743
dc.identifier.urihttp://www.kky.zcu.cz/cs/publications/ZbynekZajic_2011_Initializationof
dc.identifier.urihttp://hdl.handle.net/11025/16951
dc.description.abstractfMLLR si v porovnání s jinými adaptačními metodami vystačí s malým počtem adaptačních dat. Nicméně extrémně malé množství může vyústit ve špatný odhat adaptační matice. Taková situace se předchází dodáním nějaké informace a-priory. V tomto článku diskutujeme možnost inicializace pomocí statistik od podobných řečníků.cs
dc.format8 s.cs
dc.format.mimetypeapplication/pdf
dc.language.isoenen
dc.publisherSpringercs
dc.rights© Zbyněk Zajíc - Lukáš Machlica - Luděk Müllercs
dc.subjectfMLLRcs
dc.subjectadaptacecs
dc.subjectstatistikycs
dc.subjectrozpoznávání řečics
dc.subjectrobustnostcs
dc.subjectinitializacecs
dc.titleInitialization of fMLLR with sufficient statistics from similar speakersen
dc.title.alternativeInicializace fMLLR statistikami od podobných řečníkůcs
dc.typečlánekcs
dc.typearticleen
dc.rights.accessopenAccessen
dc.type.versionpublishedVersionen
dc.description.abstract-translatedOne of the most utilized adaptation techniques is the feature Maximum Likelihood Linear Regression (fMLLR). In comparison with other adaptation methods the number of free parameters to be estimated significantly decreases. Thus, the method is well suited for situations with small amount of adaptation data. However, fMLLR still fails in situations with extremely small data sets. Such situations can be solved through proper initialization of fMLLR estimation adding some a-priori information. In this paper a novel approach is proposed solving the problem of fMLLR initialization involving statistics from speakers acoustically close to the speaker to be adapted. Proposed initialization suitably substitutes missing adaptation data with similar data from a training database, fMLLR estimation becomes well-conditioned, and the accuracy of the recognition system increases even in situations with extremely small data sets.en
dc.subject.translatedfMLLRen
dc.subject.translatedadaptationen
dc.subject.translatedstatisticsen
dc.subject.translatedspeech recognitionen
dc.subject.translatedrobustnessen
dc.subject.translatedinitializationen
dc.type.statusPeer-revieweden
Appears in Collections:Články / Articles (KKY)
Články / Articles (NTIS)

Files in This Item:
File Description SizeFormat 
ZbynekZajic_2011_Initializationof.pdfPlný text222,99 kBAdobe PDFView/Open
Show simple item record


Please use this identifier to cite or link to this item: http://hdl.handle.net/11025/16951

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

search
navigation
  1. DSpace at University of West Bohemia
  2. Publikační činnost / Publications
  3. Fakulta aplikovaných věd / Faculty of Applied Sciences
  4. NTIS - Nové technologie pro informační společnost / NTIS - New technologies for the Information Society
  5. Články / Articles (NTIS)