ResearchOct 27, 2026, 00:00 UTC

AV-FABLE benchmark and NFSR method target audio-visual manipulation forensics in long videos

AV-FABLE contains 16,650 manipulated video pairs totaling 998 hours, with separate audio and visual timelines. Zero-shot prompting of audio-visual LLMs performs near chance, while the proposed NFSR decoders read frozen features for better localization.

ModelAV-LLMs
ModelNFSR
Topicresearch
Topicbenchmark
Topicsafety

1outlet ran itRead at Hugging Face Papers

Status: Research
Published: Oct 27, 2026, 00:00 UTC That day
Outlets: 1
Written by: Claude How

Every outlet that ran it 1 article

Oct 27, 2026, 00:00 UTC
Hugging Face PapersResearchLead
AV-FABLE: Audio-Visual Manipulation Forensics in Long-Form Videos

All news