Emotion Recognition using Voice with Machine Learning Model - Android App

This paper presents a real-time emotion recognition system that categorizes human emotions from voice inputs using machine learning, which has been integrated into an Android app. The main objective is to provide non-intrusive emotional evaluations for a range of applications, from mental health monitoring to human-computer interaction. The framework uses acoustic features such as pitch, energy levels, and Mel-Frequency Cepstral Coefficients (MFCCs), and Random Forest and Support Vector Machine (SVM) classifiers to determine nine emotional states: neutral, calm, happy, sad, angry, excited, fearful, disgusted, and surprised. The model, after training, is deployed through a Flask API on a cloud service, thus enabling smooth communication with the mobile app. Experimental results show a general accuracy rate of 87%, proving the efficiency and suitability of the framework for real-world, mobile-based emotion recognition applications.