Development of a Multimodal Framework for Deepfake Detection: Combining Visual and Audio Analysis
Bibliographic record
Abstract
Machine learning and social media advancements enable the rapid spread of realistic fake content, encompassing images, videos, and audio.Initially, fake content generation primarily focused on manipulating either audio or video streams.However, recent advancements in deep learning have enabled more sophisticated alterations, commonly called "deepfakes."While existing research predominantly concentrates on detecting fake videos by exploiting either visual or audio modalities, few approaches address audio-visual deep-fake detection.Nevertheless, these methods often need more accuracy when evaluated on a multimodal dataset with deepfake videos and manipulations in both streams.Due to neglecting facial features in preprocessing and using traditional training models.In response to this challenge, we propose a robust audio-visual deepfake detection (MAVDD) approach that analyzes audio and visual streams to enhance detection capabilities.Effectively utilizing pretrained models in image classification tasks for detecting visual deepfakes, alongside advanced preprocessing techniques for optimal facial and audio features extraction.Our experiments conducted on the multimodal audio-visual deepfake dataset "FakeAVCeleb" demonstrate that our proposed approach surpasses both unimodal (audio-only and visual-only) and multimodal (audio-visual) deepfake detection approaches in terms of accuracy and AUC (Area Under The Curve) as dedicated to tables I, II, and III.The implementation of our research work and the dataset are publicly available at the following link: https://github.com/mutlimodalDeepfakeDetection/AV-detector.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".