Classification of Thai consonant naming using Thai tone

This paper proposes the novel technique for separation of Thai consonant naming or consonant spelling using its tones. Consonant spelling is used for many applications such as a voice-actuated typewriter that helping to correct the confusable word in sound. Because fundamental frequency (F0) can be suitably used in tone classification for Thai speech recognition, which is tonal language of five patterns: mid, low, falling and rising. Consequently, classification of Thai consonant naming algorithm used F0 for distinguishing rising tone from mid tone. From the experiment result, we found that not only the level of F0 indicates tonality of sound but also considering flattening, rising, oscillation and continuity of F0 that are necessary for Thai tonal language. This paper shows performance of algorithm that classifies 996 sounds and yielding 1.72% error-rate.