An investigation of the subjective quality of non-linear loudspeaker models employing Volterra series approximations

Abstract

This investigation is an extension of the work of Angelo Farina on the use of Volterra-series approximations to model acoustic systems. Farina's method uses a logarithmically swept sine chirp to extract the transfer function of a system, including its harmonic nonlinearities. I have used this approach to construct computer models of several loudspeakers. I then evaluated my models using a blind listening test to compare their outputs against recordings of the actual loudspeakers. The results of the blind test suggest that, while this algorithm is not sufficient to convincingly model the loudspeaker transfer function, it might be useful as part of a multi-stage algorithm.