Audio Cross Verification Using Dual Alignment Likelihood Ratio Test
2026-07-20 • Sound
Sound
AI summaryⓘ
The authors explore a way to check if short viral videos of news audio have been tampered with by comparing the audio clip directly to a trusted original news recording. Instead of looking for signs of editing inside the clip itself, they compare it to the original to see if they match. They do this by testing two possibilities: that the clip is either untouched or altered, finding the best way to line up both versions, and then comparing how likely each scenario is. Their method is faster and more reliable than older techniques and helps explain why a clip is considered genuine or fake. This provides a new tool that works alongside existing methods for spotting audio tampering.
audio tampering detectionviral videoscross verificationlikelihood ratio testalignmentMFCC featuresEuclidean distanceexternal consistencytrusted source
Authors
Heidi Lei, Arm Wonghirundacha, Irmak Bukey, TJ Tsai
Abstract
This paper explores a way to verify that audio has not been maliciously tampered in a specific context: short viral videos taken from news recordings. Rather than trying to detect artifacts of tampering (internal inconsistency), we focus on positively verifying a query against a trusted source such as a news recording (external consistency). We propose a method for cross verifying a short audio query against a reference recording from which it was taken. Our approach is to define two hypotheses (non-tampered vs tampered), calculate the most likely alignment between query and reference for each hypothesis, and then perform a likelihood ratio test on the two alignments. We show that this method is fast to compute, much more robust than using MFCC features with Euclidean distance, and has the key benefit of explainability. Our cross verification approach provides an alternative perspective and complementary tool to existing tampering detection methods.