In this paper, we present a modified version of the Learning Vector Quantization (LVQ) called Dynamic Vector Quantization (DVQ). We compare the performances of both classifiers based on competitive learning on speech classification tasks. All the experiments clearly highlight the ability of generalization of the DVQ algorithm, it gives best result than LVQ2 on the test set. Moreover, DVQ always provides substancial gain in memory size and consequently is less time- consuming. Keywords: classification, vector quantization, evaluation, artificial neural network, acoustic-phonetic decoding, speech recognition.
Cite as: Poirier, F. (1991) DVQ: dynamic vector quantization application to speech processing. Proc. 2nd European Conference on Speech Communication and Technology (Eurospeech 1991), 1003-1006, doi: 10.21437/Eurospeech.1991-240
@inproceedings{poirier91_eurospeech,
author={Franck Poirier},
title={{DVQ: dynamic vector quantization application to speech processing}},
year=1991,
booktitle={Proc. 2nd European Conference on Speech Communication and Technology (Eurospeech 1991)},
pages={1003--1006},
doi={10.21437/Eurospeech.1991-240}
}