Skip to content

Dataset · j-shen3

MAVOS-DD multilingual deepfake benchmark

j-shen3/MAVOS-DD

Multilingual open audio-video deepfake benchmark with train and validation splits.

What's inside

MAVOS-DD is a multilingual audio-video deepfake benchmark supporting multiple languages (ar, ro, en, de, hi, es). It includes metadata for filtering subsets via the datasets library and supports open-set evaluation scenarios.

What you can do with it

  • Training and evaluating deepfake detectors
  • Testing open-set multimodal forgery detection models
Tasks
Video classification
Languages
Arabic, Romanian, English, German, Hindi, Spanish, Russian
Size
10K–100K rows
Downloads this month
119.7K
Likes
0

How to load it

from datasets import load_dataset

ds = load_dataset("j-shen3/MAVOS-DD")