In this document, different noise reduction algorithms based on Multichannel Wiener Filter (MWF) are described. Additionally, a method based on a derivation of MWF-N technique is proposed. The proposed method modifies the MWF-N, by adapting the trade-off parameter μ in each frequency bin, in order to increase the improvement in noise reduction without introducing large distortion on the speech. The proposed method is compared to other state-of-the-art MWF techniques. All the methods are implemented in Matlab and their performance is evaluated by three objective metrics related to speech quality, noise reduction level and speech intelligibility. The results show that the proposed method has a higher noise reduction without significantly degrading speech quality, outperforming the other techniques under the evaluated scenario.