Nonlinear System Identification for Horn Loudspeakers

For small amplitudes the dominant nonlinear mechanisms in horn loudspeakers can be modeled by a special class of polynomial systems known as S M systems. This model is tested on a real horn loudspeaker, and unknown model parameters are determined by system identification. A novel identification technique is presented which is based on the partial measurement of the higher order system functions with a steady-state excitation signal. The amplitude and phase responses of the linear subsystems in the S M system are estimated from the system functions by linear regression. The results of the system identification give qualitative clues about the dominant nonlinearities as well as quantitative estimates of relevant parameters of the loudspeaker model. This information is valuable for improving the design of horn loudspeakers and for developing nonlinear correction filters to compensate for the nonlinearities in the loudspeaker and thus reduce the distortion in the radiated sound by preprocessing the electric input signal.