基于改进EMD的语音增强方法研究

EMMD can solve the endpoint effect problem exits in EMD, but existing music noise, in the signal. Therefore, in this paper proposes a speech enhancement method based on EMMD/MMSE. This algorithm make full use of the characteristics that MMSE can solve the music noise issues exits in EMMD signal processing, so as to obtain good speech enhancement performance. Through doing experiments, comparing the original and enhancement SNR, and subjective judgment, we reach a conclusion that the speech enhancement based on EMMD/MMSE is superior to the traditional enhancement algorithm in improving the speech signal SNR and intelligibility.