Emotion Recognition using Speech Processing

Speech Emotion Recognition (SER) is a growing field in artificial intelligence focused on identifying human emotions from vocal input. This research presents a web-based system that processes speech signals, extracts meaningful acoustic features, and applies a machine learning model to classify emotions such as happy, sad, angry, and neutral. The study utilises the RAVDESS dataset [1], and features are extracted using Librosa [2]. The model, trained with a Multi-Layer Perceptron (MLP), achieves a test accuracy of 72.22% [3]. A Flask-based web application is built for emotion analysis through uploaded audio files. The system has potential applications in mental health monitoring, interactive assistants, and human-computer communication.