Phoneme-Based Speech Segmentation using Hybrid Soft Computing Framework