Speech enhancement based on Discrete Cosine Transform
Rubo Zhang · Harbin Gongcheng Daxue Xuebao/Journal of Harbin Engineering University · 2007
When the speech signal and the random noise are transformed with discrete cosine transform,the DCT coefficients in the low frequency regions are predominantly speech,and at the high frequency regions are predominantly noise.According to this,a speech enhancement method of threshold based on discrete cosine transform is proposed.The level of the trace of the signal presented in the high frequency region of the DCT coefficients is estimated to obtain the threshold parameter,then after thresholding the coefficients,the algorithm is improved by using signal-to-noise ratio estimation and speech present probability.The experiment shows that it is an effective speech enhanced method.