Method for adaptive echo compensation

摘要

The method involves using an echo compensator with a finite impulse response (FIR) filter (5). The filter reproduces the expected echoes, which is then subtracted from the affected signal. The filter coefficients (c1,...cN) are determined according to a normalised least mean square algorithm (NLMS algorithm). The step width (alpha) in the NLMS algorithm is varied according to signal source (1) surrounding conditions, echo compensator efficiency and interference level. A normalising signal for the filter is determined from the largest coefficient and then for all coefficients so that coefficient values are dynamically scaled. The shortest signal path time from the source to a sink (2) is determined according to largest coefficient temporal distance. From this the effective range of the echo compensator is determined. This information is used to estimate the remaining rest coupling.