Adaptive Equalization for Acoustic Stereo Effect

Mark Colgan

Mentor: Ken Jenkins

 

            Signal processing techniques can be used to create the apparent effect of 3-D sound from a single closely-spaced set of loudspeakers.  It has been shown that adaptive and inverse filters in combination allow crosstalk cancellation and the ability to correct for the changes in orientation of a listener’s head.  Published results have consistently utilized planar arrangements of speakers to provide 3-D sound without vertical components.

            This research investigates how alterations in the signal processing algorithms and the physical speaker geometries affect frequency response, sound quality, and range of a listener’s motion.  Utilizing the Matlab programming environment, the filters will be modeled and a sound input will be processed.  The computer’s sound output will be connected to an amplifier and the speaker array.  The system is trained at the positions of the listener’s ears by using a pair of microphones connected to the filtering system.  After the system has been initialized to the acoustical environment, the microphones may be removed. 

            The primary focus of this research is determining an ideal physical orientation of the speakers to create accurate 3-D sound including vertical components.  In this process, adaptations to the filters will be analyzed and documented as new geometries are developed and tested.