Hindi is syllabic in nature, syllable is good choice for speech synthesis. Hence, in this paper focus has been on identification of syllable units. Rule based algorithm is applied on phonetic word to identify syllables in words. With initial experiments showing that syllable split through rule has promising results. But, in practical applications, there can be loan words from other languages which will reduce accuracy. Our algorithm tested over high frequency 6000 Hindi words. It has given accuracy up to 75%. Accuracy of syllable splitter depends upon schwa deletion. If schwa deletion gives wrong output then the syllable splitter also shows wrong result. If word is formed by joining two different root words or attaching prefix or suffix then schwa deletion gets wrong and word is split inaccurately.
Volume 11 | 05-Special Issue
Pages: 1744-1748