Please use this identifier to cite or link to this item: http://hdl.handle.net/11422/2568
Type: Relatório
Title: Automatic speech recognition: a comparative evaluation between neural networks and hidden markov models
Author(s)/Inventor(s): Thomé, Antonio Carlos Gay
Diniz, Suelaine dos Santos
Santos, Sidney Cerqueira Bispo dos
Silva, Dirceu Gonzaga da
Abstract: In this work we do a comparative evaluation between Artificial Neural Networks (RNA's) and Continuous Hidden Markov Models (CDHMM), in the framework of the recognition of isolated words, under the constrain of using a small number of features extracted from each voice signal. In order to accomplish such comparison we used two models of neural networks: the Multilayer Perceptron (MLP) and a variant of the Radial Basis (RBF), and some HMM models. We evaluated the performance of all models using two different test set and observed that the neural models presented the best results in both cases. Seeking to improve the HMM performance we developed a hybrid system, HMM/MLP, that improved the results previously obtained with all HMMs, and even those obtained with the neural networks for the all previous HMM, and even the neural nets for the hardest test set case.
Keywords: Reconhecimento automático de voz
Redes neurais (Ciência da computação)
Modelos markovianos
Subject CNPq: CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAO
Department : Instituto Tércio Pacitti de Aplicações e Pesquisas Computacionais
In: Relatório Técnico NCE
Issue: 1499
Issue Date: 31-Dec-1999
Publisher country: Brasil
Language: eng
Right access: Acesso Aberto
Citation: THOMÉ, C. G. T. et al. Automatic speech recognition: a comparative evaluation between neural networks and hidden markov models. Rio de Janeiro: NCE, UFRJ, 1999. 4 p. (Relatório Técnico, 14/99)
URI: http://hdl.handle.net/11422/2568
Appears in Collections:Relatórios Técnicos e de Pesquisa

Files in This Item:
File Description SizeFormat 
14_99_000611273.pdf726,22 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.