Experimental Results on Isolated Speech Recognition System

Gavril TodereanMarieta GâtaAlexandru Căruntu



Full PDF


In this paper we described the development of Romanian ANN speech recognition system, which uses artificial neural networks. The network, which is composed by a three layer (a Multilayer Perceptron), is trained by conventional Back-propagation algorithm. The ANN speech recognition system based on Mel Frequency Cepstral Coefficients was developed using Matlab toolkit. The system was tested on the 50 isolated word Romanian speech corpus and it was training on a 50 distinct isolated word recognition task. The word recognition accuracy obtained was about 76%.

Additional Information


Alexandru Căruntu, Gavril Toderean, Marieta Gâta