东北大学学报(自然科学版) ›› 2011, Vol. 32 ›› Issue (11): 1554-1557.DOI: -

• 论著 • 上一篇    下一篇

基于EEMD和ICA的语音去噪方法

李晶皎;安冬;王骄;   

  1. 东北大学信息科学与工程学院;
  • 收稿日期:2013-06-19 修回日期:2013-06-19 发布日期:2013-04-04
  • 通讯作者: -
  • 作者简介:-
  • 基金资助:
    国家自然科学基金资助项目(60970157);;

Speech denoising method based on the EEMD and ICA approaches

Li, Jing-Jiao (1); An, Dong (1); Wang, Jiao (1)   

  1. (1) School of Information Science and Engineering, Northeastern University, Shenyang 110819, China
  • Received:2013-06-19 Revised:2013-06-19 Published:2013-04-04
  • Contact: An, D.
  • About author:-
  • Supported by:
    -

摘要: 语音去噪技术是语音识别系统走向实用化的一个关键性难题.针对语音信号为非平稳信号的特点,提出了一种基于EEMD和ICA相结合的语音去噪方法,首先利用集合经验模态分解(EEMD)算法将含噪语音信号分解为若干个独立的固有模态函数(IMF),消除了经验模态分解(EMD)算法处理语音信号时产生的模态混迭现象;然后将固有模态函数通过改进的独立分量分析(ICA)算法分离出若干个有效的语音信号分量;最后对其进行语音重构,从而达到消除噪声干扰的目的.实验结果表明,该方法在输入信噪比为-10dB的汽车噪声条件下,可以将语音信号的信噪比提高到2.741 2 dB.

关键词: 经验模态分解, 集合经验模态分解, 固有模态函数, 独立分量分析, 语音去噪

Abstract: Speech denoising technology is one of the key problems in the practical application of speech recognition systems. Since speech signals are nonstationary, speech signal contained chirp was decomposed into several intrinsic mode functions (IMF) with the method of ensemble empirical mode decomposition (EEMD). At the same time, it eliminated the model mix superposition phenomenon which usually came out in processing speech signal with the algorithm of empirical mode decomposition (EMD). After that, several effective speech signal components were separated from intrinsic mode function through the algorithm of improved independent component analysis (ICA). Finally, reconstructed them in the purpose of noise elimination. The result showed that the new speech denoising method proposed above improves SNR up to 2.7412 dB in the condition that -10 dB SNR vehicle interior noise.

中图分类号: