In this paper, a novel two-element-microphone-array-based speechenhancement algorithm is proposed. This algorithm is designed toachieve better overall performance with relatively small array size. Afrequency domain adaptive null-forming is used, in which adaptive noisecancelation is implemented in auditory subbands. And an OM-LSA basedpostfiltering stage further purifies the output. The algorithm alsofeatures interaction between the array processing and the postfilter tomake the filter adaptation more robust. This approach achievesconsiderable improvement on Signal-to-noise ratio (SNR) and subjectivequality of the desired speech. Experiments confirm the effectiveness ofthe proposed system.