5th International Conference on Spoken Language Processing

Sydney, Australia
November 30 - December 4, 1998

Evaluation and Integration of Neural-Network Training Techniques for Continuous Digit Recognition

John-Paul Hosom (1), Ronald A. Cole (1), Piero Cosi (2)

(1) Oregon Graduate Institute of Science and Technology (OGI), USA
(2) Institute of Phonetics -- C. N. R., Italy

This paper describes a set of experiments on neural-network training and search techniques that, when combined, have resulted in a 54% reduction in error on the continuous digits recognition task. The best system had word-level accuracy of 97.52% on a test set of the OGI 30K Numbers corpus, which contains naturally-produced continuous digit strings recorded over telephone channels. Experiments investigated effects of the feature set, the amount of data used for training, the type of context-dependent categories to be recognized, the values for duration limits, and the type of grammar. The experiments indicate that the grammar and duration limits had a greater effect on recognition accuracy than the output categories, cepstral features, or a 50% increase in the amount of training data.

Full Paper

Bibliographic reference.  Hosom, John-Paul / Cole, Ronald A. / Cosi, Piero (1998): "Evaluation and integration of neural-network training techniques for continuous digit recognition", In ICSLP-1998, paper 0613.