On robust time delay estimation in room acoustic environments

Abstract

Time delay estimation, which serves as the first stage that feeds into subsequent processors for localizing and tracking radiating sources, has been an active research topic for decades. This paper deals with time delay estimation in room acoustic environments. The major focus is on developing algorithms that can achieve robust time delay estimates in reverberant and noisy environments. Three approaches are investigated, i.e., the blind channel identification based method, the multichannel cross correlation based technique, and a method based on the use of the Householder transformation. The first approach achieves its robustness with respect to reverberation by taking into account reverberation in the problem formulation; the second one improves the robustness by exploiting the redundancy provided by multiple microphones, and the third method attempts to separate noise and reverberation from the signal components. Experimental results are presented to illustrate their performance and pros and cons in room acoustic environments where reverberation and noise are commonly encountered.