Combining linear and non-linear processing in the time and in the spectral domain for non-stationary noise filtering

Through this paper, the combination of linear andnon-linear techniques for noise filtering is proposed.A first linear processing stage in the time domain (anadaptive lattice ladder filter)[2],[5] is followed by anon-linear processing block in the frequency domain.The association of both procedures provides a muchhigher level of cancellation than their individualapplication. An added result is that the application ofthe frequency domain non-linear processing makes itpossible to use very short adaptive filters, with theresulting savings in computational powerrequirements.

[1]  S. Haykin,et al.  Adaptive Filter Theory , 1986 .

[2]  V. Rodellar,et al.  Implementation of an adaptive noise canceller on the TMS320C31-50 for non-stationary environments , 1997, Proceedings of 13th International Conference on Digital Signal Processing.

[3]  John G. Proakis,et al.  Digital Communications , 1983 .

[4]  María Victoria Rodellar Biarge,et al.  Dynamic adjustment of the forgetting factor in adaptive filters for non-stationary noise cancellation in speech , 1998, Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '98 (Cat. No.98CH36181).

[5]  Simon Haykin,et al.  Adaptive filter theory (2nd ed.) , 1991 .

[6]  B. Widrow,et al.  Adaptive noise cancelling: Principles and applications , 1975 .

[7]  S. Boll,et al.  Suppression of acoustic noise in speech using spectral subtraction , 1979 .