A speech segmentation system that classifies audio into speech (foreground) and background using machine learning. It supports both Least Squares and MLP classifiers, includes feature extraction, model training, frame-wise prediction, and evaluation against ground truth annotations.