Linear and nonlinear predictive analysis by iterative operation for speech signals

摘要

In this paper, an effective scheme to improve the accuracy of adaptive prediction is presented. Convergence characteristics of two adaptive algorithms; one is linear and the other is nonlinear, are investigated at first. And it is shown that by implementing the adaptive prediction in an iterative fashion for the analysis frame, the performance of the predictor is significantly improved. Some properties of the iterative analysis method are confirmed by simulation experiments on real speech data, and the superiority of the method is shown.