中文说明：处理信号的时频分析，实现了对10个数字音的识别，进行逐步线性回归，构成不同频率的调制信号，滤波求和方式实现宽带波束形成，通过反复训练模板能有较高的识别率。
English Description：
Time-frequency analysis for signal processing, realizes 10 digital voice recognition, stepwise linear regression, constitutes a different frequency modulation, filter band beam forming the sum through practice templates can have a higher recognition rate.