HalluAudio: A Comprehensive Benchmark for Hallucination Detection in Large Audio-Language Models

2026-04-21

AI Summary

This work addresses the prevalent issue of acoustically unfounded hallucinations in large audio language models (LALMs) across speech, environmental sound, and music tasks, exacerbated by the lack of comprehensive evaluation benchmarks in scale, modality coverage, and diagnostic depth. To bridge this gap, we introduce HalluAudio, the first large-scale benchmark for audio hallucination detection, comprising over 5,000 human-verified question-answer pairs designed to systematically elicit and assess hallucinations through adversarial prompts, mixed-audio inputs, and diverse task formats—including binary classification and open-ended questioning. We propose a fine-grained evaluation framework featuring metrics such as hallucination rate, yes/no bias, error typology, and refusal rate, enabling unified cross-modal assessment across the three major audio domains. Experiments on both open- and closed-source LALMs reveal critical deficiencies in acoustic grounding, temporal reasoning, and musical comprehension, laying a robust foundation for improving LALM reliability.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Cognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

Large Audio-Language Models (LALMs) have recently achieved strong performance across various audio-centric tasks. However, hallucination, where models generate responses that are semantically incorrect or acoustically unsupported, remains largely underexplored in the audio domain. Existing hallucination benchmarks mainly focus on text or vision, while the few audio-oriented studies are limited in scale, modality coverage, and diagnostic depth. We therefore introduce HalluAudio, the first large-scale benchmark for evaluating hallucinations across speech, environmental sound, and music. HalluAudio comprises over 5K human-verified QA pairs and spans diverse task types, including binary judgments, multi-choice reasoning, attribute verification, and open-ended QA. To systematically induce hallucinations, we design adversarial prompts and mixed-audio conditions. Beyond accuracy, our evaluation protocol measures hallucination rate, yes/no bias, error-type analysis, and refusal rate, enabling a fine-grained analysis of LALM failure modes. We benchmark a broad range of open-source and proprietary models, providing the first large-scale comparison across speech, sound, and music. Our results reveal significant deficiencies in acoustic grounding, temporal reasoning, and music attribute understanding, underscoring the need for reliable and robust LALMs.

Problem

• hallucination
• Large Audio-Language Models
• benchmark
• audio grounding
• model evaluation

Innovation

• hallucination detection
• large audio-language models
• multimodal benchmark
• adversarial prompting
• acoustic grounding