Understanding Sounds, Missing the Questions: The Challenge of Object Hallucination in Large Audio-Language Models

TL;DR

This paper investigates the reliability of large audio-language models, revealing their limitations in understanding discriminative questions about object sounds, and explores prompt engineering to improve their performance.

Contribution

The study introduces methods to assess object hallucination in LALMs and evaluates their performance on discriminative audio questions, highlighting key weaknesses and potential improvements.

Findings

LALMs are comparable to specialized models in understanding audio content.

LALMs struggle with discriminative questions about object sounds.

Prompt engineering can enhance LALMs' performance on discriminative tasks.

Abstract

Large audio-language models (LALMs) enhance traditional large language models by integrating audio perception capabilities, allowing them to tackle audio-related tasks. Previous research has primarily focused on assessing the performance of LALMs across various tasks, yet overlooking their reliability, particularly concerning issues like object hallucination. In our study, we introduce methods to assess the extent of object hallucination of publicly available LALMs. Our findings reveal that LALMs are comparable to specialized audio captioning models in their understanding of audio content, but struggle to answer discriminative questions, specifically those requiring the identification of the presence of particular object sounds within an audio clip. This limitation highlights a critical weakness in current LALMs: their inadequate understanding of discriminative queries. Moreover, we explore the potential of prompt engineering to enhance LALMs' performance on discriminative questions.