一种基于自适应模糊滤波的语音增强方法

在语音识别和语者识别中，通常需要先将输入的语音信号进行去噪处理，这样可使识别的正确率大大提高，通常采用基于LMS算法和RLS算法的自适应线性滤波器来进行去噪。提出了一种基于自适应模糊滤波器的语音增强方法，该模糊滤波器是一种非线性滤波器，它在语音信号的特征域空间采用参数映射的方式来滤除噪声，并能够进行自适应结构调整和参数更新。实验结果表明，采用自适应模糊滤波器来滤除噪声比线性滤波器具有更好的效果。

A New Method of Speech Enhancement Based on Adaptive Fuzzy Filter

In speech recognition and speaker recognition, the input speech signal often need to denoise in order to improve the recognition accuracy, the adaptive linear filters based on LMS or RLS algorithm are often used in filtering noise. A solution of enhancing speech by adaptive fuzzy filter is proposed, the fuzzy filter is a kind of nonlinear filters, which filters the noise by parameter mapping in the feature domain of speech signal, and it can also optimize its structure and parameters adaptively. Experiment results show that the performance of the fuzzy filter is better than that of linear filters in reducing noise.