通过改变想要处理的帧，和预测器阶数，对原始语音帧和预测语音帧作一个比较，并作出短时谱和lpc谱，最后作倒谱比较。-By changing the processing of the desired frame, and the predictor order, the original speech frame and the predicted speech frame to make a comparison and to make short-term spectrum and the l

For restoration of time-domain signals, an estimate of the instantaneous magnitude spectrum is combined with the phase of the noisy signal, and then transformed via an inverse discrete Fourier transform to the time domain. In terms of comput