这里介绍了一种基于自适应滤波的噪声抵消法，采用归一化最小均方误差算法，采集实际噪声环境下各种不同信噪比的带噪语音样本进行降噪处理，实验结果表明，处理后信号的信噪比得到了较大程度的提高，大大改善了听音效果，具有很高的可懂度，且语音自然度好，没有失真；并与谱减法进行了比较，自适应噪声抵消法的降噪幅度比谱减法有一定提高，在听音效果上，用自适应噪声抵消法处理后的语音在清晰度，自然度方面优于谱减法。-Here a novel adaptive noise cancellation method using normalized least mean square algorithm, collecting a variety of actual noise environment, the SNR of noisy speech samples for noise reduction, experimental results show that treatment After the signal to noise ratio was greater improvement in the level, greatly improved the listening results, with high intelligibility and naturalness of speech is good, no distortion　and were compared with the spectral subtraction, adaptive noise cancellation law than the spectral subtraction noise reduction rate of a certain improvement in the listening results, and with the adaptive noise cancellation method in the treatment of voice clarity, naturalness is better than spectral subtraction.