对原始语音帧进行处理，然后进行预测，最后显示出倒谱之类的东西-The original speech frame processing, and then forecast, and finally show that the class of the like