一种用于降低回响的自适应算法

An Adaptive Algorithm for Reducing Reverberation

目的回响是影响自适应噪声抵消器 (abapitvenoisecanceller,ANC)性能的重要因素 ,本文拟提出一种降低回响的改进算法。方法根据剩余均方误差随着有用信号功率的增加而增加的思想 ,在S .Ikeda算法的基础上 ,采用改进的算法。结果在 4种不同输入信噪比情况下 ,改进算法的输出信噪比比S .Ikeda算法分别提高了 5 .4dB、2 .8dB、0 .4dB和 0 .3dB。且输入信噪比在 0dB时改进算法的MOS值高于S .Ikeda算法的MOS值。结论和S .Ikeda算法相比较 。

Objective To present a modified algorithm for reducing reverberation. The reverberation is an important factor influencing the performance of an adaptive noise canceller(ANC). Method Since the excess mean squared error increases with desired signal power, a modified algorithm was proposed on the basis of S.Ikeda algorithm. Result Output SNRs of the modified algorithm increased by as much as 5.4 dB,2.8 dB, 0.4 dB and 0.3 dB respectively in comparison with those of S.Ikeda algorithm under four input SNR′s and the mean opinion scores for the modified algorithm was higher than the scores for S.Ikeda algorithm when input SNR was 0 dB. Conclusion Comparing with S.Ikeda algorithm, the modified algorithm could reduce the reverberation to a considerable degree.