Tone Classification for Isolated Thai Words using Multi-Layer Perceptron

This paper proposed a tone classification for isolated Thai words using Multi-Layer Perceptron (MLP). According to the syllable structure effect of initial consonants, vowels, and final consonants on tone recognition, three tone feature set were extracted from F-0 contour to capture the characteristic of five Thai tones and feed it into the MLP to classify five different tones of target corresponding to five Thai tones. Two dataset of Thai monosyllable were used for training and testing. The data were collected from 5 male and 5 female of Thai native speakers. The results show that these tone feature set which extracted from F-0 can be achieved a high performance of Thai tone recognition.