Deeflyzer: Hybrid Model to Detect Complex Deepfake in Digital Media

DOI

10.2991/978-94-6463-738-0_41How to use a DOI?

Keywords

Deepfake; InceptionV3; Wav2Vec 2.0; LSTM; Hybrid-Model

Abstract

The credibility of digital media has been significantly threatened due to advanced Artificial Intelligence techniques which are generally known as Deepfake, endanger the authenticity of audio-visual media. This paper presents an enhanced approach for identifying deepfake. This project utilizes a detection method that compares the area of manipulated face and surrounding areas with InceptionV3 and LSTM which are Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) respectively. Our experiments showcased that Wav2Vec 2.0 proved to be optimal for audio spoofing and artificially generated voice. The feature of localization of the fabricated media is implemented to visualize the manipulated portions. For video, datasets like DFDC, FaceForensics++ and Celeb-DF were used to train the model out of which DFDC proved to be compatible. For audio, ASVSpoof 2019, ADD 2022, FAD etc. datasets were used for training and it was clear that the ASVSpoof 2019 was the most suitable. The combination of Inception V3 and LSTM gave the accuracy of 96.17% and Wav2Vec 2.0 would offer a solution with strong potential for high accuracy of 96% in detecting sophisticated audio deepfakes. This method leverages a novel integration of CNN, RNN architecture and achieves state-of-the-art performance in detecting audio and video deepfake.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Soham Kolapkar
AU  - Riya Kshirsagar
AU  - Charudatta Thakare
AU  - Shantanu Shinde
AU  - Jitendra Musale
PY  - 2025
DA  - 2025/06/22
TI  - Deeflyzer: Hybrid Model to Detect Complex Deepfake in Digital Media
BT  - Proceedings of the International Conference on Advances and Applications in Artificial Intelligence (ICAAAI 2025)
PB  - Atlantis Press
SP  - 504
EP  - 520
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6463-738-0_41
DO  - 10.2991/978-94-6463-738-0_41
ID  - Kolapkar2025
ER  -