This paper describes the implementation of a large vocabulary speaker independent continuous speech recognition system for the DARPA RM task. The continuous HMM approach is used with mel-cepstra plus time differenced mel-cepstra and a new subword HMM topology. A set of within-word context dependent phone units are generated from a simple context merging scheme by counting and trained by both Baum-Welch and Segmental K-means algorithms. Only 6 continuous mixture diagonal Gaussian PDFs are used in each model. The recognition algorithm is based on a frame synchronous Viterbi beam search, which produces a lattice of words during forward searching, and N-best sentence candidates can be extracted in the reverse tracking, subject to the different levels of knowledge constraints. On the 991-word DARPA RM task, we present the preliminary results on the Feb-89 test set and the original SPHINX test set.
Cite as: Song, J.M., Thomas, T., Patel, M. (1991) Experiments of 991-word speaker independent continuous speech recognition on DARPA RM task. Proc. 2nd European Conference on Speech Communication and Technology (Eurospeech 1991), 643-646, doi: 10.21437/Eurospeech.1991-158
@inproceedings{song91_eurospeech,
author={J. M. Song and T. Thomas and M. Patel},
title={{Experiments of 991-word speaker independent continuous speech recognition on DARPA RM task}},
year=1991,
booktitle={Proc. 2nd European Conference on Speech Communication and Technology (Eurospeech 1991)},
pages={643--646},
doi={10.21437/Eurospeech.1991-158}
}