Microphone arrays are powerful tools for noise suppression in a reverberant room. Generalized Sidelobe Canceller (GSC) that exploits Minimum Variance (MV) criterion is efficient in interference suppression when there is no correlation between the desired signal and the interferences. Correlation between the desired signal and any of interference produces a desired signal cancellation and degradation of signal-to-noise ratio. This paper analyses the unwanted cancellation of the desired source. It shows that cancellation level of the desired signal is proportional to the correlation between the direct wave and the reflected waves. For prevention of a desired signal cancellation we suggest the GSC parameter estimation during the pauses of the desired signal. For this case it is analytically shown that there is no cancellation of the desired signal. The proposed algorithm was experimentally tested and compared with the Conventional Beamformer (CBF) and GSC. Experimental tests have shown the advantage of the proposed method.