Robust Splitting: Uses Minimum Energy + Zero Crossing detection to handle fast speech without glitches.