The purpose of sound field reproduction is to use loudspeakers to produce desired sound at particular locations in a given environment, which has a wide range of applications such as virtual reality, etc. The computational load required to design and implement filters involved in sound reproduction systems can be significant, especially when the desired sound has rich information over a wide frequency band. To reduce the computational load, sub-band filtering approaches are usually used in sound reproduction systems. In the present work, an approach is proposed to design the sub-band filters used in sound reproduction systems in a more convenient way, where the filter design problem is formulated into a convex optimization problem. Detailed analysis has been conducted on how to specify the response characteristics of each sub-band and how different sub-band filters can be combined into one full band filter in the design and implementation of the system. Results also show that even if the sub-band filter structure is not necessary, this approach can also be applied to reduce the computational load in designing inverse filters when the plant responses involve relatively large differences in delay time among different frequency bands.
Sub-band filter, Sound field reproduction, Convex optimization, Cone programming, reduced order technique
Acoustics and Noise Control, Controls and Electromechanical Systems
Date of this Version