Abstract: To improve the performance of sound source localization under coherent background noise,the theory of strong tracking filter is combined with the array-data acquisition model,to propose an algorithm for the fast estimation of phase shift.This algorithm introduces multiple suboptimal fading factors,which can automatically adjust with the noise phase mutation to ensure the output residual sequence is orthogonal to each other,and the useful information is further extracted.The continuous phase-mutation tracking performance is verified by numerical simulations,the phase shift is accurately estimated under the condition of parameter mismatch,and the performance is stable in a wide frequency range.The results of broadband noise are given by speaker experiments.The interferences can be removed on the fly,and the convergence speed is more than doubled compared with the existing algorithm.This research has achieved accurate estimation of random phase shift,and improves the real-time sound source localization capability of the acoustic array under the interference of coherent noise.