非特定人普通话孤立数字音识别系统

A Speaker Independent Isolated Mandarin Digit Recognition System

通过分析汉语语音及非特定人语音特点，选择过零率、能量、基音周期和前两个共振峰作为语音特征，用音素分类的方法构造判决树，在此基础上，给出了一个非特定人普通话孤立数字音（０～９）识别系统，识别率大约为９６．６％。

Based on an analysis of the characteristics of Mandarin digits,the zero-crossing rate,energy,pitch and the first two formant frequencies are chosen as the features in the speakerindependent,isolated Mandarin digit recognition system. The recognition accuracy of about96.6%is achieved in the experiments,The system is different from most similar ones inthat it recognizes Mandarin digits by a decision tree rather than the pattern matchingmethod.The decision tree first classifies the Mandarin digits into two types,the voiced consonant type (0,1,2,5,6, 8) and the unvoiced one(3,4,7, 9)and then it gives the recognition result by the first two formant frequencies and the pitches. It is shown in the experi-ments that the decision tree method derived from the phonetic classification is very effectivefor the recognition of Mandarin digits.