ISCA Archive ICSLP 1994
ISCA Archive ICSLP 1994

Recurrent neural network word models for small vocabulary speech recognition

Philippe Le Cerf, Dirk Van Compernolle

In this paper, we describe the use of recurrent Multilayer Perceptrons (MLP's) as state probability estimators for word models. The advantage over conventional word Hidden Markov Models (HMM's) is the ease of discriminative training of the models. We find that a minimal state duration is useful. The results on a telephone quality, speaker independent digit recognition task compare favorably with the results of the approach presented by us earlier this year (Le Cerf et al [4]).


Cite as: Cerf, P.L., Compernolle, D.V. (1994) Recurrent neural network word models for small vocabulary speech recognition. Proc. 3rd International Conference on Spoken Language Processing (ICSLP 1994), 1547-1550

@inproceedings{cerf94_icslp,
  author={Philippe Le Cerf and Dirk Van Compernolle},
  title={{Recurrent neural network word models for small vocabulary speech recognition}},
  year=1994,
  booktitle={Proc. 3rd International Conference on Spoken Language Processing (ICSLP 1994)},
  pages={1547--1550}
}