The Wiener filter is used to estimate a desired signal by filtering an observed noisy signal using linear time-invariant filtering, assuming known stationary signal and noise spectra with additive noise. The Wiener filter coefficients are calculated to minimize the average squared distance between the filter output and desired signal. The input signal is convolved with the Wiener filter and compared to a reference signal to obtain the filtering error, with the output being an estimate of noise or interference subtracted from the desired signal.