Truthlens: Ai-powered Deepfake Detection Using Deep Learning And Media Feature Analysis

Abstract

The Rapid Advancement Of Artificial Intelligence Has Made It Possible To Generate Highly Convincing Manipulated Media, Commonly Referred To As Deepfakes. These Synthetic Videos And Audio Recordings Pose A Serious Threat To Information Integrity, Personal Reputation, And Public Trust. TruthLens Is An AI-powered Deepfake Detection System Developed To Help Users Verify The Authenticity Of Uploaded Video And Audio Files. The System Is Built Using A MERN Stack Web Application Integrated With A Python Flask Backend Responsible For Running Deep Learning Inference. For Video Analysis, The System Employs MTCNN For Face Detection And A Transformer-based Deepfake Classification Model To Analyse Sampled Frames. Audio Analysis Uses Spectral Feature Extraction Techniques To Identify Patterns Associated With Synthetic Speech. Experimental Results Show That The System Accurately Classifies Media As Real Or Fake With A Meaningful Confidence Score. This Work Demonstrates How Machine Learning And Neural Network Techniques Can Be Deployed In Full-stack Applications To Address The Growing Challenge Of Digital Misinformation.

Keywords