摘要： 为提高重构语音质量,提出一种子空间语音增强算法。在内嵌式预白化子空间方法的基础上,用特征值置换增强语音效果。将纯净语音和噪声的协方差矩阵进行广义特征值分解后,大特征值分量包含纯净语音信息,而小特征值分量包含噪声信息,待特征值排序后,用相邻的大特征值分量置换小特征值分量,可有效提高语音帧的相关性,获得更好的增强语音信号。相比传统的子空间方法,该算法适用于强噪声环境,能显著提高信噪比。

关键词: 语音质量, 语音增强, 子空间方法, 特征值分解, 语音帧

Abstract: A subspace speech enhancement approach is proposed for a better reconstructed speech quality.On the basis of Generalized Eigenvalue Decomposition(GEVD) subspace approach for speech enhancement,speech quality is improved by eigenvalue substitution.The study shows that after the generalized eigenvalue decomposition for the covariance matrices of the clean speech signal and the noise,the larger eigenvalue component contains the main information of speech signal,while the smaller eigenvalue component contains noise information.After the eigenvalue ordering,the smaller eigenvalue and its corresponding eigenvectors are replaced with the larger ones so that there is a better enhanced speech signal and the correlation of the speech frame can be improved greatly.Compared with other subspace methods,the proposed method can restrain strong noise more effectively,and get a higher signal noise ratio.

Key words: speech quality, speech enhancement, subspace approach, eigenvalue decomposition, speech frame

中图分类号: 

• TN912.3