This project uses deep learning to recognize emotions from speech. Based on EMO-DB and RAVDESS datasets, it extracts audio features (MFCCs) to classify emotions like anger, joy, and sadness, enabling applications in voice assistants and mental health.