Result Details

Study of linear transformations applied to training of cross-domain adapted large vocabulary continuous speech recognition systems

KARAFIÁT, M. Study of linear transformations applied to training of cross-domain adapted large vocabulary continuous speech recognition systems. Brno: 2009. 73 p.
Type
dissertation
Language
English
Authors
Karafiát Martin, Ing., Ph.D., FIT (FIT), DCGM (FIT)
Abstract

This thesis investigates into two important issues of acoustic modeling for automatic speech recognition (ASR).  The second part deals with using heterogeneous data resources in ASR training.

Keywords

LVCSR system, meeting recognition, linear transform, Adaptation, cross domain adaptation, HLDA, CMLLR, MLLR, narrow band - wide band

URL
Published
2009
Pages
73
Place
Brno
BibTeX 
@misc{BUT192664,
  author="Martin {Karafiát}",
  title="Study of linear transformations applied to training of cross-domain adapted large vocabulary continuous speech recognition systems",
  year="2009",
  pages="73",
  address="Brno",
  url="http://www.fit.vutbr.cz/research/groups/speech/publi/2009/karafiat_thesis.pdf"
}
Projects
Augmented Multi-party Interaction, EU, Sixth Framework programme, 506811-AMI, start: 2004-01-01, end: 2006-12-31, completed
Speech Recognition under Real-World Conditions, GACR, Standardní projekty, GA102/08/0707, start: 2008-01-01, end: 2011-12-31, completed
Research groups
Výzkumná skupina dolování dat z řeči BUT Speech@FIT (RG SPEECH)
Departments
Ústav počítačové grafiky a multimédií (DCGM)
Back to top