The problem of noise suppression by exploiting microphone array (MA) has long been an attractive direction of speech enhancement. Recently, most studies have been devoted to MA beamformer, which aim at preserving the target speech component while suppressing all the background noise and interference. Dual - microphone arrays is one of the most useful MA geometry to obtain the desired target speaker while suppressing all background noise without speech distortion due to the flexible combination of many observed characteristics. However, in realistic recording environment, MA beamforming’s performance often degraded or increased because of many unwanted reasons, such as: the error of the MA distribution, the imprecise direction of arrival (DoA) of target talker, the different MA sensitivities, the presence of third-party talker. In order to work well in all situations, this contribution presents an approach of exploiting phase to derive a reliable Wiener filter. The illustrated experiments have confirmed the effectiveness of suggested methods to extract desired speech components while eliminating surroundings. The results show the advantage of noise reduction 18.6 (dB), increasing the speech quality in the term of signal-to-noise ratio (SNR) from 12.6 to 14.5 (dB) in practical environments. Furthermore, the above technique is possible to apply post processing filtering in various MA systems with a large microphone.