ISCA Archive ICSLP 2000
ISCA Archive ICSLP 2000

Speech recognition using error spotting

Rachida El Méliani, Douglas O'Shaughnessy

Spontaneous conversational phone-call speech databases are dicult to recognize because of the large variation of speech rates, of pronunciations as well as noises, of acoustic degradations from the telephone channel, and of an unpredictable non-grammatical language structure including many random phenomena. Each cause of misrecognition can be addressed separately; however there is still no satisfying solution. As a misrecognition is considered by the system as being a kind of new word, we propose to apply here our keyword spotting and new-word detection design. However because of the large variety of types of misrecognitions and of the lack of information on where, why and how they occur, we had to define a di erent language model than those used in previous work. Results show for most of the processed files a noticeable improvement in the recognition rate, often associated with a decrease in the number of substitutions and a slight increase in the number of the deletions.

Cite as: El Méliani, R., O'Shaughnessy, D. (2000) Speech recognition using error spotting. Proc. 6th International Conference on Spoken Language Processing (ICSLP 2000), vol. 2, 1057-1060

  author={Rachida {El Méliani} and Douglas O'Shaughnessy},
  title={{Speech recognition using error spotting}},
  booktitle={Proc. 6th International Conference on Spoken Language Processing (ICSLP 2000)},
  pages={vol. 2, 1057-1060}