Exploring Audio Hallucination in Egocentric Video Understanding

TL;DR

This paper investigates audio hallucinations in egocentric video understanding, revealing that current AV-LLMs often generate inaccurate sounds, and introduces an evaluation framework and dataset to analyze and quantify these hallucinations.

Contribution

It presents a systematic evaluation framework, a curated dataset, and a taxonomy for analyzing audio hallucinations in egocentric videos, highlighting the unreliability of current AV-LLMs.

Findings

Advanced AV-LLMs like Qwen2.5 Omni have high hallucination rates.

Models achieve only 27.3% accuracy on foreground sound questions.

Models achieve only 39.5% accuracy on background sound questions.

Abstract

Egocentric videos provide a distinctive setting in which sound serves as crucial cues to understand user activities and surroundings, particularly when visual information is unstable or occluded due to continuous camera movement. State-of-the-art large audio-visual language models (AV-LLMs) can generate multimodal descriptions. However, we show in this work that they are prone to audio hallucinations, often inferring sounds from visual cues that are visible but not heard. We present a systematic and automatic evaluation framework for analyzing audio hallucinations in egocentric video through a targeted question-answering (Q/A) protocol. We curate a dataset of 300 egocentric videos and design 1,000 sound-focused questions to probe model outputs. To characterize hallucinations, we propose a grounded taxonomy that distinguishes between foreground action sounds from the user activities and background ambient sounds. Our evaluation shows that advanced AV-LLMs, such as Qwen2.5 Omni, exhibit high hallucination rates, achieving only 27.3% and 39.5% accuracy on Q/As related to foreground and background sounds, respectively. With this work, we highlight the need to measure the reliability of multimodal responses, emphasizing that robust evaluation of hallucinations is essential to develop reliable AV-LLMs.