The acoustic emission/microseismic technique (AE/MS) has emerged as one of the most important techniques in recent decades and has found wide applications in different fields. Extraction of seismic event with precise timing is the first step and also the foundation for processing AE/MS signals. However, this process remains a challenging task for most AE/MS applications. The process has generally been performed by human analysts. However, manual processing is time consuming and subjective. These challenges continue to provide motivation for the search for new and innovative ways to improve the signal processing needs of the AE/MS technique. This research has developed a highly efficient method to resolve the problems of background noise and outburst activities characteristic of AE/MS data to enhance the picking of P-phase onset time. The method is a hybrid technique, comprising the characteristic function (CF), high order statistics, stationary discrete wavelet transform (SDWT), and a phase association theory. The performance of the algorithm has been evaluated with data from a coal mine and a 3-D concrete pile laboratory experiment. The accuracy of picking was found to be highly dependent on the choice of wavelet function, the decomposition scale, CF, and window size. The performance of the algorithm has been compared with that of a human expert and the following pickers: the short-term average to long-term average (STA/LTA), the Baer and Kradolfer, the modified energy ratio, and the short-term to long-term kurtosis. The results show that the proposed method has better picking accuracy (84% to 78% based on data from a coal mine) than the STA/LTA. The introduction of the phase association theory and the SDWT method in this research provided a novelty, which has not been seen in any of the previous algorithms.