Context-Dependent Multiple Distribution Phonetic Modeling with MLPs
Abstract
A number of hybrid multilayer perceptron (MLP)/hidden Markov model (HMM:) speech recognition systems have been developed in recent years (Morgan and Bourlard. 1990). In this paper. we present a new MLP architecture and training algorithm which allows the modeling of context-dependent phonetic classes in a hybrid MLP/HMM: framework. The new training procedure smooths MLPs trained at different degrees of context dependence in order to obtain a robust estimate of the cootext-dependent probabilities. Tests with the DARPA Resomce Management database have shown substantial advantages of the context-dependent MLPs over earlier cootext(cid:173) independent MLPs. and have shown substantial advantages of this hybrid approach over a pure HMM approach.
Cite
Text
Cohen et al. "Context-Dependent Multiple Distribution Phonetic Modeling with MLPs." Neural Information Processing Systems, 1992.Markdown
[Cohen et al. "Context-Dependent Multiple Distribution Phonetic Modeling with MLPs." Neural Information Processing Systems, 1992.](https://mlanthology.org/neurips/1992/cohen1992neurips-contextdependent/)BibTeX
@inproceedings{cohen1992neurips-contextdependent,
title = {{Context-Dependent Multiple Distribution Phonetic Modeling with MLPs}},
author = {Cohen, Michael and Franco, Horacio and Morgan, Nelson and Rumelhart, David E. and Abrash, Victor},
booktitle = {Neural Information Processing Systems},
year = {1992},
pages = {649-657},
url = {https://mlanthology.org/neurips/1992/cohen1992neurips-contextdependent/}
}