Published August 7, 2026 | Version v2

A Modular Heuristic Framework for Multimodal Synthetic Media Analysis

Authors/Creators

  • 1. ROR icon GIFT University

Description

This preprint presents a modular heuristic framework for multimodal synthetic media analysis, designed to detect deepfakes and AI-generated content across five modalities: text, audio, image, video, and lip-sync. The framework integrates modality-specific detection heuristics into a unified system, addressing the growing challenge of synthetic media forensics. A working implementation of the system is deployed and publicly accessible on HuggingFace Spaces (aneela-pervez/DEEP-FAKE-DETECTOR).

Files

12_Multimodal_Synthetic_Media_Analysis_Pervez.pdf

Files (747.4 kB)

Name Size
md5:94fd50c0f5b72a6b7f13731f56d37d70
747.4 kB Preview Download

Additional details

Dates

Updated
2026-06-03

Software

Repository URL
https://huggingface.co/spaces/aneela-pervez/DEEP-FAKE-DETECTOR
Programming language
Python
Development Status
Active