计算机工程与应用 ›› 2010, Vol. 46 ›› Issue (17): 139-141.DOI: 10.3778/j.issn.1002-8331.2010.17.039

• 数据库、信号与信息处理 • 上一篇    下一篇

基于Hilbert-Huang变换的语音增强技术研究

夏敏磊1,2,徐 振1,俞祁焰3,范影乐3   

  1. 1.浙江机电职业技术学院 电子信息系,杭州 310056
    2.浙江大学 电气工程学院,杭州 310027
    3.杭州电子科技大学 自动化学院,杭州 310018
  • 收稿日期:2008-11-26 修回日期:2009-02-02 出版日期:2010-06-11 发布日期:2010-06-11
  • 通讯作者: 夏敏磊

Study on speech enhancement based on Hilbert-Huang transform

XIA Min-lei1,2,XU Zhen1,YU Qi-yan3,FAN Ying-le3   

  1. 1.Zhejiang Institute of Mechanical & Electrical Engineering,Hangzhou 310056,China
    2.College of Electrical Engineering,Zhejiang University,Hangzhou 310027,China
    3.College of Automation,Hangzhou Dianzi University,Hangzhou 310018,China
  • Received:2008-11-26 Revised:2009-02-02 Online:2010-06-11 Published:2010-06-11
  • Contact: XIA Min-lei

摘要: 提出了一种基于Hilbert-Huang变换的语音增强方法。首先利用经验模态分解方法(Empirical Mode Decomposition,EMD),选择合适的固有模态函数对含噪语音进行初步降噪,然后根据信噪比分别确定过减因子进行谱减运算。实验结果表明,与传统的谱减法相比,该方法的输出信噪比提高了5 dB以上,尤其在非稳定噪声条件下,输出性能有更为明显的改善。经Hilbert-Huang变换后得到的特征量能较为有效地描述语音信号的非线性以及非平稳特性。

关键词: Hilbert-Huang变换, 语音增强, 谱减法

Abstract: A speech enhancement method based on Hilbert-Huang transform is presented.The empirical mode decomposition method is applied to denoise initially,and then the spectral subtraction method is used to enhance each intrinsic mode function.The subtraction factor has been justified due to the variation of the SNR.The results of enhancement experiments show that,this method outperforms the standard power spectral subtraction method resulting in superior speech quality,especially in non-stationary noise environment.Hilbert-Huang method gives the true description of the non-linear and non-stationary characteristics of speech signals,and it has wide application prospect.

Key words: Hilbert-Huang transform, speech enhancement, spectral subtraction

中图分类号: