一种语音信号的线性局部自适应递推预测算法

Adaptive Local Linear Prediction of Speech Signals

语音产生过程包含非线性过程,传统的线性预测方法不能很好地解决这些非线性成份。局部线性预测是一种高精度的预测算法,但计算复杂度较大。为提高非线性预测的速度,提出了一种自适应递推局部线性预测算法,并设计算法的步骤,分析算法的复杂性。仿真结果表明,该算法比线性预测算法精度高,是一种有效的语音信号非线性预测方法。

Speech production includes nonlinear process, but the nonlinear part in the speech signal is not taken into account in the linear models and is not solved well with the traditional linear prediction. Local linear prediction is a high precise algorithm, but its application to the speech signals is impeded by its prohibitive computational complexity. In order to overcome this problem,an adaptive local linear prediction is presented. Its implement steps, the analysis of the computational complexity are also given. Comparative simulation between the speech with local linear prediction and linear prediction shows that adaptive recursive local linear prediction of speech is an efficient prediction of speech.