So they need statistical techniques to match the pairs. Even in this process there are multiple variations. Since syllable boundaries are difficult to find and join, they seek to cut a syllable in the middle and join with a like complementary cut syllable in the adjoining syllable pair. The other thing i find is in English they seek to extract pairs of syllables, because some approaches to speech synthesis work with concatenation of syllables. While english speech analysis seeks to capture thousands of words to extract syllables, Indian alphabets themselves appear to be a comprehensive set of syllables. they can be easier and simpler because Indian languages are more phonetic. It has been my opinion that Indian languages are different from English for speech processing.
0 kommentar(er)
