GCC-PHAT time difference estimation algorithm based on binary frequency weight with suppressing wind noise

Abstract

Aiming at the problems that GCC-PHAT algorithm is sensitive to additive noise and the weighted GCC-PHAT algorithm based on prior SNR can′t eliminate the jamming of non-stationary noise——wind noise, an improved GCC-PHAT algorithm is presented.The improved algorithm utilizes the binary frequency weight calculated by magnitude of coherence and correlation in signal adjacent frames to eliminate frequency component disturbed by wind noise. Meanwhile, computational load of the algorithm would be modulated with voice activity according to binary weight. The experiment results show that the proposed algorithm can effectively suppress wind noise and significantly improve computational load.

Keywords

• GCC; TDOA; binary frequency weight; wind noise