MVAD: A Benchmark Dataset for Multimodal AI-Generated Video-Audio Detection

TL;DR

The paper introduces MVAD, a comprehensive benchmark dataset for detecting AI-generated multimodal video-audio content, addressing the lack of diverse, high-quality datasets beyond facial deepfakes.

Contribution

It presents the first dataset specifically designed for multimodal AI-generated video-audio detection, covering diverse styles, content, and forgery patterns.

Findings

Dataset includes samples from various generative models.

High perceptual quality across multiple visual styles.

Diverse content categories and multimodal data types.

Abstract

The rapid advancement of AI-generated multimodal video-audio content has raised significant concerns regarding information security and content authenticity. Existing synthetic video datasets predominantly focus on the visual modality alone, while the few incorporating audio are largely confined to facial deepfakes--a limitation that fails to address the expanding landscape of general multimodal AI-generated content and substantially impedes the development of trustworthy detection systems. To bridge this critical gap, we introduce the Multimodal Video-Audio Dataset (MVAD), the first comprehensive dataset specifically designed for detecting AI-generated multimodal video-audio content. Our dataset exhibits three key characteristics: (1) genuine multimodality with samples generated according to three realistic video-audio forgery patterns; (2) high perceptual quality achieved through diverse state-of-the-art generative models; and (3) comprehensive diversity spanning realistic and anime visual styles, four content categories (humans, animals, objects, and scenes), and four video-audio multimodal data types. Our dataset will be available at https://github.com/HuMengXue0104/MVAD.