The video shows on an example made of synthetic data, the use of the Wiener filter to transform an autoregressive signal of order 2 into a Dirac's impulsion. The animation shows from top to bottom:

• the source signal which is an autoregressive signal of order 2, which duration can be changed by modifying, the overvoltage coefficient of the filter (by moving cursor);
• the desired signal which is a Dirac's impulsion, which temporal position can be changed by moving cursor;
• the output signal which corresponds to the estimation of the Dirac's impulsion by the Wiener filter. It is possible to choose a causal operator or a non-causal operator.

If the operator is causal and if the desired signal is a Dirac's impulsion, the computation of the Wiener operator only requires the computation of the source signal's autocorrelation. The output signal is an approximation of a Dirac's impulsion, situated at the source instant (arrival time) of the source signal. If the operator is non-causal and if the desired signal is a Dirac's impulsion, the computation of the Wiener operator requires the computation of the autocorrelation of the source signal and the computation of the cross-correlation between the source signal and the desired signal. The output signal is an approximation of a Dirac's impulsion, situated at the temporal position of the Dirac's impulsion which is the desired signal.