Parallel adaptive filter structures for acoustic noise cancellation

The structure of an adaptive filter is shown to have significant effects on its tracking performance in a nonstationary environment. The cancellation of additive acoustic automobile noise is used to compare the performance of parallel adaptive filter structures, with the results showing that a structure consisting of a two-stage canceler using cancellation on subfrequency bands provides the best performance with a maximum noise spectrum reduction of 28 dB and a total average noise power reduction of 16 dB. The reference microphone adaptive noise canceler approach proposed by B. Widrow et al. (1975) can be extended through the inclusion of subbanded filters to focus the taps on particular subbands of the noise signals and also through the use of multiple reference microphones to provide references for the various noise sources in the multireference, multisource environment to improve the noise reduction performance.<
>