Syllable segmentation of Chinese continuous speech based on combination knowledge

Syllable segmentation is very important for Chinese continuous speech recognition. There is special structure in Chinese speech syllable which consists of SHENMU and YUNMU. In this paper, all kinds of conditions for Chinese speech syllable connections were listed, on the basis, some rules were proposed, all the rules were combined to use segmentation in our system. By using the general parameters E, ZCR and related parameters and the proposed rules, syllable segmentation algorithm of continuous speech mixed with noise was designed and implemented.