Predicting Malay Prominent Syllable Using Support Vector Machine

Abstract

Abstract The prediction of prominent syllable using Support Vector Machines (SVMs) is proposed in this paper. As much as 50 sentences out of 400 sentences from Malay syntax-prosody speech text has been selected as training and testing data set. SVM was trained to classify the prominent syllable in the sentences according to the feature presented. A prominent word can be categorized as prominent syllables but the prominent syllable has the potential not to classify as prominent word. Using training data, we trained it to obtain a model and the model will be used to predict instances with testing data set. The kernel function namely Radial Basis Function (RBF) is used to find the best cross validation accuracy of training data with parameter, C and γ. The SVM technique also compared with a Naive Bayes (NB) classifier and the comparison clearly claims that the proposed technique based architecture outperforms on all experiments. Using SVM classifier, the percentage accuracy of the classification archived rate of 88.7304% while 88.3024% using the Naive Bayes (NB). The results showed that SVM is conveniently classified because the performance shown is better than NB.