Modeling the precedence effect for sound localization in reverberant environment
Jie Huang, Noboru Ohnishi, Noboru Sugie · 2002
This paper presents a model-based method for sound localization of concurrent continuous speech sources in reverberant environment. The model of the precedence effect was proposed to detect echo-free onsets by specifying a typical impulse response pattern. Fine structure time differences, calculated by zero-crossing method, were integrated into an azimuth histogram by restrictions between three microphones. Two sound sources were localized in both an anechoic chamber and a normal room with strong reverberations. The time segment needed for localization was 0.5 to 2 seconds and the accuracy was about a few degrees in both environments.