UNIGE document Chapitre d'actes
previous document  unige:2265  next document
add to browser collection
Title

Comparing Rule-Based and Statistical Approaches to Speech Understanding in a Limited Domain Speech Translation System

Authors
Rayner, Manny
Hockey, Beth Ann
Chatzichrisafis, Nikos
Published in TMI 2004 - 10th International Conference on Theoretical and Methodological Issues in Machine Translation. Baltimore (Maryland, USA). 2004, p. 21-29
Abstract The paper directly compares two versions of a medical speech translation system, one with a grammar based language model (GLM) recognizer and the other with a statistical language model (SLM) recognizer. We construct the GLM using a corpus-based method, so that both the GLM and the SLM can be derived from the same corpus; evaluation is carried out with respect to performance on the speech translation task. Despite using a very small training set for both the GLM and the SLM, the SLM delivers much better word error rates on unseen test material. Nonetheless, evaluating both systems on translation performance rather than word error rates, the GLM-based version of the system outperforms the SLM on the actual translation task.
Full text
Proceedings Chapter (106 Kb) - public document Free access
Structures
Research group TIM/ISSCO
Citation
(ISO format)
RAYNER, Manny et al. Comparing Rule-Based and Statistical Approaches to Speech Understanding in a Limited Domain Speech Translation System. In: TMI 2004 - 10th International Conference on Theoretical and Methodological Issues in Machine Translation. Baltimore (Maryland, USA). [s.l.] : [s.n.], 2004. p. 21-29. https://archive-ouverte.unige.ch/unige:2265

220 hits

224 downloads

Update

Deposited on : 2009-07-29

Export document
Format :
Citation style :