Spatio-Temporal Adaptive Inverse Filtering in the Wave Domain

Abstract

The sound quality for acoustical communication, information, and entertainment is often subject to impairments by room reflections and undesired noise sources. As a remedy, various signal processing techniques have been developed for different applications, like acoustic echo cancelation and active noise control. Starting from the single channel case, these techniques have recently been extended to multiple channels. These extensions increased the intricacy of the original problem by the added complexity of the multichannel case. The effect was that the resulting techniques like multichannel active listening room compensation and multichannel active noise control appeared as unrelated solutions to different kinds of problems. This chapter gives a unifying description of these spatio-temporal adaptive methods on the perspective of sound reproduction by tracing them back to the fundamental problem of inverse filtering. After analyzing the problems of an adaptive solution, eigenspace adaptive filtering is introduced as a concept for decoupling the multichannel problem. Unfortunately, this concept is not straightforward applicable in its pure form, since it requires data-dependent transformations. Therefore, an approximate solution called wave-domain adaptive filtering is introduced. It has the advantage of being data independent and still performs a close-to-ideal decoupling. Based on the unifying inverse filtering description, the application of wave-domain adaptive filtering to active listening room compensation and to active noise control is shown.