Integrated direct sub-band adaptive volterra filter and its application to identification of loudspeaker nonlinearity

In this paper, we propose a novel realization of sub-band adaptive Volterra filter, which consists of input signal transformation block and only one adaptive Volterra filter. The proposed realization can focus on major frequency band, in which a target nonlinear system has dominant components, by changing the number of taps in each sub-band in order to simultaneously realize high computational efficiency and high identification performance. The proposed realization of sub-band adaptive Volterra filter is applied to the identification of electro-dynamic loudspeaker systems and the effectiveness is demonstrated through some simulations. Simulation results show that the proposed realization can significantly improve the estimation accuracy.