In this paper, we present a new method based on wavelet transform for moment tensor inversion. The method consists of a semi-automatic data preparation and source inversion that is prepared in python. The method is similar to the point source technique in the time domain but is adjusted using wavelet coefficients. The advantage of our algorithm is to test the centroid moment tensor in all frequency content using lower samples compared to the time domain. We show the results of accuracy and reliability of the method to obtain the complexity of earthquake sources with synthetic tests and real observed seismograms.