Title: | Initialization of adaptation by sufficient statistics using phonetic tree |
Authors: | Zajíc, Zbyněk Machlica, Lukáš Müller, Luděk |
Issue Date: | 2012 |
Publisher: | IEEE Press |
Document type: | článek article |
URI: | http://www.kky.zcu.cz/cs/publications/ZbynekZajic_2012_Initializationof http://hdl.handle.net/11025/16949 |
Keywords: | rozpoznávání řeči;adaptace;inicializace;fonetický strom |
Keywords in different language: | speech recognition;adaptation;initialization;phonetic tree |
Abstract: | Článek popisuje řešení problému malého mknožství dat pro fMLLR adaptaci. Navrhuje inicializaci statistik z dat od nejbližších řečníků s využitím informace o fonetické příslušnosti. |
Abstract in different language: | In this work we deal with the problem of small amount of data when estimating a feature transformation for the speaker adaptation of an acoustic model. Our goal is to compensate for the lack of adaptation data by a proper initialization of transformation matrices. Methods used in such situations are described, they are based on collecting additional accumulated statistics from nearest speakers. The proposed initialization approach is based on accumulated statistics too, but it incorporates also phonetic information when selecting the “nearest” statistics. Initialization methods compensating for the absence of actual speaker’s data are tested on telephone recordings with different amounts of adaptation data. In worst situation with extremely small amount of adaptation data relative improvement of 5% is obtained. |
Rights: | © Zajíc, Zbyněkc - Lukáš Machlica - Luděk Müller |
Appears in Collections: | Články / Articles (KKY) Články / Articles (NTIS) |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
ZbynekZajic_2012_Initializationof.pdf | Plný text | 178,52 kB | Adobe PDF | View/Open |
Please use this identifier to cite or link to this item:
http://hdl.handle.net/11025/16949
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.