Coherence-based Dual Microphone Wind Noise Reduction by Wiener Filtering

In this paper, we propose a two-step method to reduce wind noise in dual microphone environments. Wind noise in outdoors recording often leads to critical degradation to the speech signal. Therefore, it is necessary to apply algorithms for reduction of wind noise. The proposed algorithm exploits the coherence of input signals and use a Wiener filter to noise frequency regions. For evaluating the proposed algorithm, we compare with existing algorithms in terms of the noise attenuation minus speech attenuation (NA-SA). Sentences from the IEEE sentence were recorded in meeting room environments and wind noise was collected in outdoors. Then, those signals are synthesized with variant SNR conditions. Proposed algorithm shows the best performance compared with other reference algorithms.