A general approach is presented to the design of inverse filters used in multichannel sound reproduction systems. The formulation of the problem is cast within a traditional least-squares format. Algorithms are presented which enable the rapid adaptation of inverse filters in response to measurements made in the reproduced sound field. Experimental results and computer simulations are used to illustrate the potential for using an inverse filter matrix to manipulate the position of stereophonic images. The possibilities suggested could be used in conjunction with existing stereophonic recordings in order to improve stereophonic imaging in situations where the relative locations of loudspeakers and listeners are constrained to be nonideal.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.