Large vocabulary decoding and confidence estimation using word posterior probabilities
- 7 November 2002
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 3, 1655-1658 vol.3
- https://doi.org/10.1109/icassp.2000.862067
Abstract
The paper investigates the estimation of word posterior probabilities based on word lattices and presents applications of these posteriors in a large vocabulary speech recognition system. A novel approach to integrating these word posterior probability distributions into a conventional Viterbi decoder is presented. The problem of the robust estimation of confidence scores from word posteriors is examined and a method based on decision trees is suggested. The effectiveness of these techniques is demonstrated on the broadcast news and the conversational telephone speech corpora where improvements both in terms of word error rate and normalised cross entropy were achieved compared to the baseline HTK evaluation systems.Keywords
This publication has 5 references indexed in Scilit:
- Word graph rescoring using confidence measuresPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- A post-processing system to yield reduced word error rates: Recognizer Output Voting Error Reduction (ROVER)Published by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- LVCSR log-likelihood ratio scoring for keyword spottingPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- The 1998 HTK system for transcription of conversational telephone speechPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1999
- Using word probabilities as confidence measuresPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1998