MAP Based Speaker Adaptation in Very Large Vocabulary Speech Recognition of Czech

DSpace Repository

Show simple item record Červa, Petr Nouza, Jan 2016-05-24 2016-05-24 2004
dc.identifier.issn 1210-2512
dc.description.abstract The paper deals with the problem of efficient adaptation of speech recognition systems to individual users. The goal is to achieve better performance in specific applications where one known speaker is expected. In our approach we adopt the MAP (Maximum A Posteriori) method for this purpose. The MAP based formulae for the adaptation of the HMM (Hidden Markov Model) parameters are described. Several alternative versions of this method have been implemented and experimentally verified in two areas, first in the isolated-word recognition (IWR) task and later also in the large vocabulary continuous speech recognition (LVCSR) system, both developed for the Czech language. The results show that the word error rate (WER) can be reduced by more than 20% for a speaker who provides tens of words (in case of IWR) or tens of sentences (in case of LVCSR) for the adaptation. Recently, we have used the described methods in the design of two practical applications: voice dictation to a PC and automatic transcription of radio and TV news. en
dc.description.sponsorship Grant Agency of the Czech Republic [102/02/0124]; MSM 242200001
dc.format text
dc.language.iso en
dc.publisher Spolecnost Pro Radioelektronicke Inzenyrstvi
dc.publisher Technická Univerzita v Liberci cs
dc.publisher Technical university of Liberec, Czech Republic en
dc.relation.ispartof Radioengineering en
dc.source j-scopus
dc.source j-wok
dc.subject Speech recognition en
dc.subject speaker adaptation en
dc.subject maximum a posteriori method en
dc.subject hidden Markov models en
dc.title MAP Based Speaker Adaptation in Very Large Vocabulary Speech Recognition of Czech en
dc.type Article
local.relation.volume 13
local.relation.issue 3
local.identifier.wok 208050000009
local.identifier.scopus 2-s2.0-84868255099
local.faculty Faculty of Mechatronics, Informatics and Interdisciplinary Studies
local.citation.spage 42
local.citation.epage 46
local.department Department of Electronics and Signal Processing
local.identifier.stag RIV/46747885:24220/04:00000110
local.access οpen
local.fulltext yes

Files in this item

This item appears in the following Collection(s)

Show simple item record

Search DSpace

Advanced Search


My Account