In this paper we consider a new way to calculate the logarithm of the Likelihood Ratio Function for Gaussian signals. This approach is based on the standard Kalman filter. Its efficiency is substantiated theoretically, and numerical examples show how such a method works in practice.