Designing an AI-Powered Virtual Trainer to Enhance Presentation Skills of Sri Lankan Undergraduates Through Real-Time Multimodal Feedback

Undergraduate students in Sri Lanka struggle with presentation skills due to performance anxiety, insufficient guidance, and lack of real-time feedback. SPEAKRAFT addresses this gap as an AI-powered virtual trainer that delivers intelligent, real-time coaching. It integrates ML models to evaluate facial expressions using OpenCV/DeepFace, vocal loudness, speech pacing, and filler words. The modular web platform employs FastAPI for AI orchestration, React/JavaScript for frontend interactivity, and Express JS/Mongo DB for backend functionality. Users engage in targeted practice sessions or full presentations, receiving granular feedback and visual analytics. The system also enables institutional applications for academic and professional communication development.