Beamforming is among the most widely used source localization techniques for MEG and EEG data in neuroscience. This project aims at porting an updated version of the five-dimensional beamformer to analyze time-frequency data in source space (Dalal et al., 2008; implemented in MNE-Python as “tf_lcmv”). The update adopts the Hilbert transform to obtain a continuous amplitude and phase time series per frequency band, in contrast to the previous method that utilizes discrete time windows and does not retain phase information. It also adds a weight normalization option to the LCMV beamformer (neural activity index and unit noise gain), which will be available for other LCMV versions as well. Furthermore, statistical testing of the Hilbert beamformer output using the Wilcoxon rank sum test will be implemented to complete the new Hilbert beamformer pipeline. Due to the size of the MEG data and intensiveness of the calculations, the code will be developed with memory efficiency and parallelization in mind. Interactive visualization of source-localized time frequency representations will complete the Hilbert beamformer routine.