基于线性预测的自适应语音增强技术

Speech Enhancement Using Adaptive Filter with Linear Predictor

提出了一种基于线性预测的自适应语音增强方法。该方法利用带噪语音的线性预测作为自适应滤波器的参考输入,使有色噪声的相关性隐含于参考输入之中,从而大大地提高了语音的信噪比。仿真实验证明,被环境噪声污染了的语音信号经自适应系统处理后,其信噪比提高了32.70dB。

This paper presents an adaptive speech enhancement method based on linear prediction. The method use prediction of input noisy signal, which include the correlation of noises, as a reference input of adaptive filter. Our simulation shows that the SNR (signal to noise ratio) of processing data has been improved obviously.