AV-Deepfake1M++: A Large-Scale Audio-Visual Deepfake Benchmark with Real-World Perturbations

Zhixi Cai; Kartik Kuckreja; Shreya Ghosh; Akanksha Chuchra; Muhammad Haris Khan; Usman Tariq; Tom Gedeon; Abhinav Dhall

doi:10.1145/3746027.3761979

AV-Deepfake1M++: A Large-Scale Audio-Visual Deepfake Benchmark with Real-World Perturbations

Zhixi Cai
, Kartik Kuckreja
, Shreya Ghosh
, Akanksha Chuchra
, Muhammad Haris Khan
, Usman Tariq
, Tom Gedeon
, Abhinav Dhall

Department of Data Science & AI

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Other › peer-review

2 Citations (Scopus)

Abstract

The rapid surge of text-to-speech and face-voice reenactment models makes video fabrication easier and highly realistic. To encounter this problem, we require datasets that rich in type of generation methods and perturbation strategy which is usually common for online videos. To this end, we propose AV-Deepfake1M++, an extension of the AV-Deepfake1M having 2 million video clips with diversified manipulation strategy and audio-visual perturbation. This paper includes the description of data generation strategies along with benchmarking of AV-Deepfake1M++ using state-of-the-art methods. We believe that this dataset will play a pivotal role in facilitating research in Deepfake domain. Based on this dataset, we host the 2025 1M-Deepfakes Detection Challenge. The challenge details, dataset and evaluation scripts are available online under a research-only license at https://deepfakes1m.github.io/2025.

Original language	English
Title of host publication	Proceedings of the 33rd ACM International Conference on Multimedia
Editors	Luca Rossetto, Stevan Rudinac, Duc-Tien Dang-Nguyen, Wen-Huang Cheng, Phoebe Chen, Jenny Benois-Pineau
Place of Publication	New York NY USA
Publisher	Association for Computing Machinery (ACM)
Pages	13686-13691
Number of pages	6
ISBN (Electronic)	9798400720352
DOIs	https://doi.org/10.1145/3746027.3761979
Publication status	Published - 2025
Event	ACM International Conference on Multimedia 2025 - Dublin, Ireland Duration: 27 Oct 2025 → 31 Oct 2025 Conference number: 33rd https://dl.acm.org/doi/proceedings/10.1145/3746027 (Proceedings) https://acmmm2025.org/ (Website)

Conference

Conference	ACM International Conference on Multimedia 2025
Abbreviated title	MM 2025
Country/Territory	Ireland
City	Dublin
Period	27/10/25 → 31/10/25
Internet address	https://dl.acm.org/doi/proceedings/10.1145/3746027 (Proceedings) https://acmmm2025.org/ (Website)

Keywords

datasets
deepfake
detection
localization

Access to Document

10.1145/3746027.3761979Licence: CC BY

759907776-oaFinal published version, 2.2 MBLicence: CC BY

Cite this

Cai, Z., Kuckreja, K., Ghosh, S., Chuchra, A., Khan, M. H., Tariq, U., Gedeon, T., & Dhall, A. (2025). AV-Deepfake1M++: A Large-Scale Audio-Visual Deepfake Benchmark with Real-World Perturbations. In L. Rossetto, S. Rudinac, D.-T. Dang-Nguyen, W.-H. Cheng, P. Chen, & J. Benois-Pineau (Eds.), Proceedings of the 33rd ACM International Conference on Multimedia (pp. 13686-13691). Association for Computing Machinery (ACM). https://doi.org/10.1145/3746027.3761979

Cai, Zhixi ; Kuckreja, Kartik ; Ghosh, Shreya et al. / AV-Deepfake1M++ : A Large-Scale Audio-Visual Deepfake Benchmark with Real-World Perturbations. Proceedings of the 33rd ACM International Conference on Multimedia. editor / Luca Rossetto ; Stevan Rudinac ; Duc-Tien Dang-Nguyen ; Wen-Huang Cheng ; Phoebe Chen ; Jenny Benois-Pineau. New York NY USA : Association for Computing Machinery (ACM), 2025. pp. 13686-13691

@inproceedings{61fe37ff67764e38a7eb533e208a7de6,

title = "AV-Deepfake1M++: A Large-Scale Audio-Visual Deepfake Benchmark with Real-World Perturbations",

abstract = "The rapid surge of text-to-speech and face-voice reenactment models makes video fabrication easier and highly realistic. To encounter this problem, we require datasets that rich in type of generation methods and perturbation strategy which is usually common for online videos. To this end, we propose AV-Deepfake1M++, an extension of the AV-Deepfake1M having 2 million video clips with diversified manipulation strategy and audio-visual perturbation. This paper includes the description of data generation strategies along with benchmarking of AV-Deepfake1M++ using state-of-the-art methods. We believe that this dataset will play a pivotal role in facilitating research in Deepfake domain. Based on this dataset, we host the 2025 1M-Deepfakes Detection Challenge. The challenge details, dataset and evaluation scripts are available online under a research-only license at https://deepfakes1m.github.io/2025.",

keywords = "datasets, deepfake, detection, localization",

author = "Zhixi Cai and Kartik Kuckreja and Shreya Ghosh and Akanksha Chuchra and Khan, \{Muhammad Haris\} and Usman Tariq and Tom Gedeon and Abhinav Dhall",

note = "Publisher Copyright: {\textcopyright} 2025 Copyright held by the owner/author(s).; ACM International Conference on Multimedia 2025, MM 2025 ; Conference date: 27-10-2025 Through 31-10-2025",

year = "2025",

doi = "10.1145/3746027.3761979",

language = "English",

pages = "13686--13691",

editor = "Luca Rossetto and Stevan Rudinac and Duc-Tien Dang-Nguyen and Wen-Huang Cheng and Phoebe Chen and Jenny Benois-Pineau",

booktitle = "Proceedings of the 33rd ACM International Conference on Multimedia",

publisher = "Association for Computing Machinery (ACM)",

address = "United States of America",

url = "https://dl.acm.org/doi/proceedings/10.1145/3746027, https://acmmm2025.org/",

}

Cai, Z, Kuckreja, K, Ghosh, S, Chuchra, A, Khan, MH, Tariq, U, Gedeon, T & Dhall, A 2025, AV-Deepfake1M++: A Large-Scale Audio-Visual Deepfake Benchmark with Real-World Perturbations. in L Rossetto, S Rudinac, D-T Dang-Nguyen, W-H Cheng, P Chen & J Benois-Pineau (eds), Proceedings of the 33rd ACM International Conference on Multimedia. Association for Computing Machinery (ACM), New York NY USA, pp. 13686-13691, ACM International Conference on Multimedia 2025, Dublin, Ireland, 27/10/25. https://doi.org/10.1145/3746027.3761979

AV-Deepfake1M++: A Large-Scale Audio-Visual Deepfake Benchmark with Real-World Perturbations. / Cai, Zhixi; Kuckreja, Kartik; Ghosh, Shreya et al.
Proceedings of the 33rd ACM International Conference on Multimedia. ed. / Luca Rossetto; Stevan Rudinac; Duc-Tien Dang-Nguyen; Wen-Huang Cheng; Phoebe Chen; Jenny Benois-Pineau. New York NY USA: Association for Computing Machinery (ACM), 2025. p. 13686-13691.

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Other › peer-review

TY - GEN

T1 - AV-Deepfake1M++

T2 - ACM International Conference on Multimedia 2025

AU - Cai, Zhixi

AU - Kuckreja, Kartik

AU - Ghosh, Shreya

AU - Chuchra, Akanksha

AU - Khan, Muhammad Haris

AU - Tariq, Usman

AU - Gedeon, Tom

AU - Dhall, Abhinav

N1 - Conference code: 33rd

PY - 2025

Y1 - 2025

N2 - The rapid surge of text-to-speech and face-voice reenactment models makes video fabrication easier and highly realistic. To encounter this problem, we require datasets that rich in type of generation methods and perturbation strategy which is usually common for online videos. To this end, we propose AV-Deepfake1M++, an extension of the AV-Deepfake1M having 2 million video clips with diversified manipulation strategy and audio-visual perturbation. This paper includes the description of data generation strategies along with benchmarking of AV-Deepfake1M++ using state-of-the-art methods. We believe that this dataset will play a pivotal role in facilitating research in Deepfake domain. Based on this dataset, we host the 2025 1M-Deepfakes Detection Challenge. The challenge details, dataset and evaluation scripts are available online under a research-only license at https://deepfakes1m.github.io/2025.

AB - The rapid surge of text-to-speech and face-voice reenactment models makes video fabrication easier and highly realistic. To encounter this problem, we require datasets that rich in type of generation methods and perturbation strategy which is usually common for online videos. To this end, we propose AV-Deepfake1M++, an extension of the AV-Deepfake1M having 2 million video clips with diversified manipulation strategy and audio-visual perturbation. This paper includes the description of data generation strategies along with benchmarking of AV-Deepfake1M++ using state-of-the-art methods. We believe that this dataset will play a pivotal role in facilitating research in Deepfake domain. Based on this dataset, we host the 2025 1M-Deepfakes Detection Challenge. The challenge details, dataset and evaluation scripts are available online under a research-only license at https://deepfakes1m.github.io/2025.

KW - datasets

KW - deepfake

KW - detection

KW - localization

UR - https://www.scopus.com/pages/publications/105024063455

U2 - 10.1145/3746027.3761979

DO - 10.1145/3746027.3761979

M3 - Conference Paper

AN - SCOPUS:105024063455

SP - 13686

EP - 13691

BT - Proceedings of the 33rd ACM International Conference on Multimedia

A2 - Rossetto, Luca

A2 - Rudinac, Stevan

A2 - Dang-Nguyen, Duc-Tien

A2 - Cheng, Wen-Huang

A2 - Chen, Phoebe

A2 - Benois-Pineau, Jenny

PB - Association for Computing Machinery (ACM)

CY - New York NY USA

Y2 - 27 October 2025 through 31 October 2025

ER -

Cai Z, Kuckreja K, Ghosh S, Chuchra A, Khan MH, Tariq U et al. AV-Deepfake1M++: A Large-Scale Audio-Visual Deepfake Benchmark with Real-World Perturbations. In Rossetto L, Rudinac S, Dang-Nguyen DT, Cheng WH, Chen P, Benois-Pineau J, editors, Proceedings of the 33rd ACM International Conference on Multimedia. New York NY USA: Association for Computing Machinery (ACM). 2025. p. 13686-13691 doi: 10.1145/3746027.3761979

AV-Deepfake1M++: A Large-Scale Audio-Visual Deepfake Benchmark with Real-World Perturbations

Abstract

Conference

Keywords

Access to Document

Other files and links

Cite this