Use of acoustic prior information for confidence measure in ASR (automatic speech recognition) applications


Creative Commons License

Mengusoglu E., Ris C.

ACOUSTICS RESEARCH LETTERS ONLINE-ARLO, vol.6, no.2, pp.92-98, 2005 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 6 Issue: 2
  • Publication Date: 2005
  • Doi Number: 10.1121/1.1843171
  • Journal Name: ACOUSTICS RESEARCH LETTERS ONLINE-ARLO
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.92-98
  • Hacettepe University Affiliated: No

Abstract

In this paper, a new acoustic confidence measure of automatic speech recognition hypothesis is proposed and it is compared to approaches proposed in the literature. This approach takes into account prior information on the acoustic model performance specific to each phoneme. The new method is tested on two types of recognition errors: the out-of-vocabulary words and the errors due to additive noise. An efficient way to interpret the raw confidence measure as a correctness prior probability is also proposed in the paper. (C) 2005 Acoustical Society of America.