对非平稳信号进行分段、截取，再对处理过后的短时信号做短时傅里叶变换，分析其频谱特性，并找到各个时间节点的频率特性，以便分析不同事件段内的声音特征，一图像的形式表达出，最终达到区分不同声音的目的。(The nonstationary signals are segmented, and then the interception, the short-time signal after processing the short-time Fourier transform, the analysis of the spectrum characteristics and frequency characteristics and find the different time node, in order to analyze the acoustic characteristics of different event period, expressed in the form of an image, and ultimately achieve the purpose of distinguishing different voices.)