Abstract

In this papcr, wc propose an approach for continuous speech recognition by spotting consonant-vowel (CV) units. The niain issues in spotting CV units are the location of anchor points and labelling the regions around these anchor points using suitable classifiers. The vowel onset points (VOPs) trave bcen used as an- chor points. The distribution capturing ability of autoassociative neural network (AA”) models is explored for detection of VOPs in continnous speech. We consider support vector machine (SVM) based classifiers due to their ability of generalisation from limited training data and also due to their inherent discriminative learn- ing. The CV spotting approach for continuous speech recognition has bccn dcmonstratcd for scntcnccs in Indian languages