Adaptive Non-Linear Prediction with Order Statistics for Speech in Impulsive Noise

摘要

In this paper, we investigate the linear prediction of speech signals in an impulsive noise environment. Both schemes of batch processing and adaptive processing are comparatively studied and it is shown that the adaptive processing scheme is basically suitable in a highly impulsive noise environment. As an extended version of the Order Statistic Least Mean Square (OSLMS) algorithm addressed by Shimamura et al., an OSLMS algorithm involving ambiguous sorting is developed. The performance of the proposed algorithm is demonstrated and it is shown that the effects of impulse noise are significantly suppressed by the proposed algorithm.