低信噪比弱信号提取算法研究

Investigation of an algorithm for extraction of weak signals with low signal-to-noise ratio

采用从强背景噪声中提取话音信号的自适应噪声抵消技术———最小均方误差(LMS)入手,对该算法的性能进行了分析,针对其收敛速度慢、提取信号频带窄的缺点,提出了改进的有动量因子的自适应最小均方误差算法(MLMS).仿真结果表明,该方法对于解决弱信号提取问题有一定效果.

Starting with an adaptive noise cancellation technique-least mean square(LMS) method-to separate voice signals from strong background noises and, specifically, analysing the performance of this algorithm to find its demerits such as sluggish convergence and narrow wave band of signal extraction, an improved algorithm of adaptive least mean square with momentum factor is presented. Simulation result show that this method exhibits a certain effectiveness for weak signal extraction.