本文介绍关于语波的线性预测原理,重点介绍自相关线性预测法和选择线性预测法。给出了应用线性预测分析进行语音识别时的基本参数及其计算公式。在共振峰分析和音调提取分析-合成系统方面的引用也扼要地提及。对极-零模型也给予关注,并指出这是今后的发展方向。 This article introduces the theory of linear prediction of speech, focusing on the autocorrelation linear prediction method and the choice of linear prediction method. The basic parameters and their formulas for speech recognition using linear predictive analysis are given. The reference to formant analysis and pitch extraction analysis - synthesis systems is also briefly mentioned. Polar - zero model also pay attention, and pointed out that this is the future direction of development.