A tone recognition of polysyllabic Chinese words using an approximation model of four tone pitch patterns

Abstract

Presents an approximate model based on the tangent function as a simple approximate method for recognition of four tones of Chinese speech. In order to carry out the segmentation of the syllabic unit, the authors propose a novel segmentation scheme, in which, on the basis of analyzing the sound linking characteristic between syllables, the phrases are classified into two groups as non-linked syllabic and linked syllabic. The pitch information may be directly employed for the former, and for the latter, two parameters, sound power and average magnitude differential function, are used. The authors carried out recognition experiments using three syllable phrases and received satisfactory results. >