【摘要】 语音降噪算法的目的是在保留语音信号质量的前提下，尽可能消除背景噪声，提高语音质量.为了进一步提升单通道最大信噪比滤波器的语音降噪性能，本文提出了一种融合帧间相关性与优化全带输出信噪比的语音降噪方法.该方法首先在传统最大信噪比滤波器中引入帧间相关性，随后对信号的相关矩阵进行特征值分解，筛选出语音信号占优的若干个子空间应用最大信噪比滤波器，而将其余噪声占优的子空间置零.通过这一优化过程，构建出基于帧间相关性的优化全带输出信噪比滤波器.仿真实验结果表明，与传统最大信噪比滤波器相比，所提出的方法显著提升了带噪信号的信噪比，进一步改善了语音质量，同时所提出的滤波器是一种更为通用的方法，传统维纳滤波器和最大信噪比滤波器均为其特例.更多还原

【Abstract】 The purpose of the speech noise reduction algorithm is to eliminate background noise as much as possible and improve speech quality.To further improve the performance of the single-channel maximum signal-to-noise ratio(SNR) filter for speech noise reduction, this paper proposes a noise reduction method that combines inter-frame correlation with full-band output SNR optimization.The method introduces inter-frame correlation into the traditional maximum SNR filter and performs generalized eigenvalue decomposition on the correlation matrices of speech and noise signals.By selecting subspaces where the speech signal dominates, the method applies the maximum SNR filter to these subspaces, while setting the other noise-dominated subspaces to zero.This optimization process results in the construction of an optimized full-band output SNR filter based on inter-frame correlation.Simulation results demonstrate that the proposed method outperforms the traditional maximum SNR filter in terms of both SNR improvement and speech quality enhancement.Furthermore, the proposed filter is more general, the traditional Wiener filter and the maximum signal-to-noise ratio filter are its special cases.更多还原