论文部分内容阅读
自适应噪声抵消方法(ANC)能有效地增强被加性噪声干扰的信号。鉴于传统的采用LMS横向滤波器(TF)的ANC(TFANC)有其缺点,本文研究了采用格形联合滤波器(LF)的ANC(LFANC),证明了它具有基本上不依赖于输入的优异收敛性能。另外,针对前人工作的不足,我们从理论上着重研究导出了LFANC的失调的解析表达式并进行了实验验证,从而得到失调随滤波器级数呈指数增长的重要结论。我们还成功地将优化步长用于多级LF,这样就能在几乎不增加运算量的前提下大大加快LF的收敛,将以上方法用于对真实含噪语音的处理,取得了较好的效果。
The adaptive noise cancellation method (ANC) can effectively enhance the signal that is disturbed by additive noise. In view of the shortcomings of the traditional ANC (TFANC) with LMS transversal filter (TF), this paper studies ANC (LFANC) using a lattice combined filter (LF), demonstrating that it has excellent performance that is substantially independent of input Convergence performance. In addition, in view of the shortcomings of previous work, we theoretically study and derive the analytic expression of the LFANC imbalance and experimentally verify it, so as to obtain the important conclusion that the imbalance increases exponentially with the filter order. We also succeeded in using the optimization steps for multilevel LF, which greatly accelerates the convergence of LF with almost no additional computations. The above method is applied to the processing of real noisy speech and the better effect.