Bayesian Confidence Scoring and Adaptation Techniques for Speech Recognition

Tae-Yoon KIM  Hanseok KO  

Publication
IEICE TRANSACTIONS on Communications   Vol.E88-B   No.4   pp.1756-1759
Publication Date: 2005/04/01
Online ISSN: 
DOI: 10.1093/ietcom/e88-b.4.1756
Print ISSN: 0916-8516
Type of Manuscript: LETTER
Category: Multimedia Systems for Communications" Multimedia Systems for Communications
Keyword: 
speech recognition,  confidence measure,  adaptation,  OOV rejection,  

Full Text: PDF(182.3KB)>>
Buy this Article




Summary: 
Bayesian combining of confidence measures is proposed for speech recognition. Bayesian combining is achieved by the estimation of joint pdf of confidence feature vector in correct and incorrect hypothesis classes. In addition, the adaptation of a confidence score using the pdf is presented. The proposed methods reduced the classification error rate by 18% from the conventional single feature based confidence scoring method in isolated word Out-of-Vocabulary rejection test.