A lightweight, real-time wake-word detection system built with PyTorch. This project processes 1-second audio snippets by converting them into mel-spectrograms, then classifies each as either a "wake ...