联合投影逼近子空间跟踪的语音增强算法

An UPAST approach for enhancing speech corrupted by colored noise

本文提出一种联合投影逼近子空间跟踪(UPAST)的语音增强算法。本算法以投影逼近子空间跟踪算法为基础, 在无需对噪声进行任何假定(白噪声或是有色噪声)或近似且不需要任何语音活动检测的前提下,以递推更新的方式得到 语音信号和噪声信号协方差矩阵同时对角化的特征向量和特征值,因而运算复杂度低,实现了有色噪声背景下语音信号的 最优估计。主观和客观测试都表明本算法要优于其它子空间增强算法。

Through the modification of the Projection Approximation Subspace Tracking （PAST） approach proposed by Yang, This paper proposes a speech enhancement algorithm based on the United Projection Approximation Subspace Tracking Algorithm （UPAST） which simultaneously diagonalizes speech and noise covariance matrices to efficiently attain the optimal estimation of speech corrupted by colored noise without any presumption or approximation of the property of noise and any voice activity detection （VAD）. Objective and subjective tests showed that UPAST was superior to other subspace-based methods.