This paper presents the application of a recurrent neural network model, the Gamma Memory Model (GMM), to the problem of speech segmentation. The Gamma Memory Model is composed of a local and constrained feedback. We first present the tests made with gamma units in the input plane and standard neurons in the hidden and output layers. Then, we present the tests made with gamma neurons in the hidden layers : the architecture of the neurons in the hidden layer is modified, a gamma neuron being composed of a standard neuron followed by a gamma unit.
Cite as: Buniet, L., Fohr, D. (1995) Continuous speech segmentation with the gamma memory model. Proc. 4th European Conference on Speech Communication and Technology (Eurospeech 1995), 1685-1688, doi: 10.21437/Eurospeech.1995-407
@inproceedings{buniet95_eurospeech,
author={Laurent Buniet and Dominique Fohr},
title={{Continuous speech segmentation with the gamma memory model}},
year=1995,
booktitle={Proc. 4th European Conference on Speech Communication and Technology (Eurospeech 1995)},
pages={1685--1688},
doi={10.21437/Eurospeech.1995-407}
}