Visual speech learning from an e-tutor via dynamic lip movement-based video segmentation and comparison.