本文根据孤立词语音识别系统的处理步骤，从语音信号的预处理开始，分别详细说明了每个过程，比如预加重、分帧加窗和端点检测。接着介绍了特征参数MFCC的原理与选取，最后介绍了DTW算法的特点-In this paper, based on the process steps of　isolated words speech recognition technology,　starting with the speech signal pre-processing, each process is described in detail such as pre-emphasis, add window frame and endpoint detection.　then the theory and selection of the feature parameters are introduced, finally, I introduced the characteristics of the DTW algorithm.