A special transducer-oriented filter structure is presented to compensate for nonlinear distortions of a horn loudspeaker by prefiltering the electric input signal. The filter has a block-oriented filter architecture derived from the results of nonlinear loudspeaker modeling and is composed of static nonlinear and dynamic linear subsystems. This structure requires less computational effort and less memory than other generic architectures and allows compensation for the modeled nonlinear mechanisms completely. In practice, simplified versions of this filter can be implemented to compensate for dominant nonlinearities only. System identification performed on a particular horn loudspeaker provides valuable information about the dominant nonlinearity and allows adjusting the free filter parameters. The implementation of an adequate filter structure for a professional horn loudspeaker in a DSP56000 is discussed and results of the distortion reduction are presented.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.