Voice signal filtering iteration method

The invention discloses a voice signal filtering iteration method, which comprises the following steps of: 1, selecting two Newton iteration methods from a third-order Newton iteration method, a fifth-order Newton iteration method and a seventh-order Newton iteration method; 2, combining a FastICA algorithm, and adopting the Newton iteration method selected in the step 1 to construct an iteration formula; and step 3, voice signal filtering is carried out by using the iterative formula in the step 2. The method can effectively improve the average signal-to-noise ratio, reduce the number of iterations, and improve the voice separation effect and separation speed of the blind source voice signal. 本发明公开了一种语音信号滤波迭代方法，包括步骤1：从3阶、5阶和7阶牛顿迭代法中选取两种牛顿迭代法；步骤2：结合FastICA算法，采用步骤1选取的牛顿迭代法构建迭代公式；步骤3：采用步骤2的迭代公式进行语音信号滤波。本发明可以有效提升平均信噪比，降低迭代次数，提高盲源语音信号的语音分离效果和分离速度。