Nowadays, due to the development of smart home and smart voice-controlled devices, the concern of noise reduction and speech enhancement are getting much attention. Basically, most signal processing algorithms include separating the noisy speech in each frame, estimating, tracking and updating noise statistics for mitigating noise. Recently, scholars and researchers are focusing on studying the space domain in the scenario of multichannel case with an appropriate microphone array (MA). MA owns the high directivity beampattern toward the speech source at a certain direction while eliminating all interference, which come from other directions. Minimum Variance Distortionless Response (MVDR) is commonly installed in most acoustics devices, such as: surveillance equipment, conferencing systems or mobile phones. Because of the different sensitivities of MA, the error of direction of arrival of interest signal or imprecise MA distribution, the overall MVDR beamformer’s evaluation often corrupted. In this paper, the authors introduce a new method for adaptive updating of the steering vector to enhance MVDR beamformer’s performance. The illustrated experiment has confirmed the advantage of suggested techniques in improving the speech quality in the term of signal-to-noise ratio (SNR)from 8.8 to 10.3 (dB), reducing speech distortion to 6.9 (dB). The proposed methods are suitable for applying into multi-channel systems for signal processing in realistic recording environments.