Chinese speech understanding system

Application of the knowledge of syntax, morphology, and semantics to acoustic analysis and semantic achievement is discussed. An acoustic analyzer which is restricted by syntax rules and morphology rules is proposed. The semantic achievement procedure based on syntactic structure analysis and semantic deep structure is used to obtain the meaning of input speech. In addition, to make the system accept fairly natural speech, a multiparameter segmentation method is described. The authors present experimental results obtained using the system. >