A method and its system for converting Chinese speech-dynamic human face image information are disclosed. It features that the method includes such steps as synchronously recording speeches and humanface images, converting the recorded data to digital image data, inputting to a computer, determining the sampling time of static Chinese video position, choosing FAP parameter set of human face animation from MPEG-4 standard, measuring the characteristic points of human face, calculating the FAP parameters' values, creating Chinese basic video position, set creating dynamic video position model and determining various parameters of the said model.