On-line convolutive blind source separation of non-stationary signals

We have shown previously that non-stationary signals recorded in a static m ulti-path environment can often be recovered by simultaneously decorrelatin g varying second order statistics. As typical sources are often moving, how ever, the multi-path channel is not static. We present here an on-line grad ient algorithm with adaptive step size in the frequency domain based on sec ond derivatives, which we refer to as multiple adaptive decorrelation (MAD) . We compared the separation performance of the proposed algorithm to its o ff-line counterpart and to another decorrelation based on-line algorithm.