Voice biometrics is one kind of physical characteristics that differs from each individual. Due to this uniqueness, voice classification is found useful in classifying speakers’ gender, mother tongue, ethnicity, emotion states, identity verification, verbal command control, and so forth. In this study, we propose a pre-processing methodology named Statistical Feature Extraction (SFX) since we want to facilitate voice classification through Data Mining Methodology. Using SFX we can faithfully remodel statistical characteristics of the time series voice data via a sequence of piecewise transform functions. We focus on the comparison of effects of various popular data mining algorithms on multiple datasets. And the new methodology is tested through simulation experiments over four representative types of human voice data, namely Female and Male, Emotional Speech, Speaker Identification and Language Recognition. The experiments yield encouraging results supporting the fact that SFX indeed produces better performance in voice classification than traditional signal processing techniques like Wavelets and LPC-to-CC.