谱减法语音增强的仿真实现：读取语音文件，产生随机白噪声，将两者联合起来，加入汉明窗，帧间重叠50 ，短时FFT，短时相位谱，频域中合成语音，短时IFFT并各帧重叠相加，去除汉明窗引起的增益，最后得到增强后的语音。-Spectral subtraction speech enhancement Simulation: reading voice files, generate random white noise, the two unite Jiaruhanming window, 50　overlap between frames, short FFT, short-term phase spectrum, frequency domain speech synthesis, short IFFT and overlap-add each frame to remove gain caused Hamming window, and finally obtain the enhanced speech.