将声道模型视为全极点模型，采用具有定点运算性质的Ｓｃｈｕｒ算法求得部分相关系数，利用ＬＰＣ对数倒谱分析求得语音识别系统中采用的待识特征量——倒谱．本系统以倒谱作为特征参数，采用加权倒谱距离求得待识量与以Ｋ－均值聚类所得的参数库之间的特征距离，利用最小距离原则进行识别．经验证，本文所采用的方法取得了较好的实验结果 The channel model is regarded as an all-pole model. The Schur algorithm with fixed-point operation is used to obtain partial correlation coefficients. The LPC log-cepstral analysis is used to find the feature quantity of celestial features used in speech recognition systems, ie cepstrum. The system uses the cepstrum as the characteristic parameter, and uses the weighted cepstrum distance to obtain the characteristic distance between the unknown quantity and the parameter library obtained by K-means clustering, and uses the minimum distance principle to identify. After verification, the method adopted in this paper has achieved good experimental results