Voice Morphing Detection System

Abstract

Voice morphing technology has advanced to a level where artificial or altered voices can closely imitate real human speech, creating serious challenges in authentication, security, and communication. This project proposes a Voice Morphing Detection System that identifies whether an input voice is genuine or morphed using digital signal processing and machine learning techniques. The system extracts key audio features such as MFCC, spectral properties, pitch, jitter, and formants, which are then analyzed by a trained ML classifier to detect manipulation patterns present in synthesized or modified speech. To support real-time monitoring, the project integrates an IoT-based audio acquisition module using an ESP32 microcontroller and a microphone sensor. The IoT device captures live voice data and transmits it wirelessly to the server or cloud for processing. This improves system accessibility and enables remote detection in applications like call verification, identity authentication, customer service, and fraud prevention. The proposed system provides an efficient, portable, and scalable solution capable of distinguishing natural human voice from artificially morphed voice with high accuracy. By combining IoT hardware with intelligent audio analysis, the project contributes to enhancing security against voice spoofing, impersonation, and AI-generated audio threats.