ReasonAudio: A Benchmark for Evaluating Reasoning Beyond Matching in Text-Audio Retrieval

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Honglei, Chen, Yuting, Hu, Chenpeng, Zhang, Siyue, Shi, Yilei
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910194084085760
author Zhang, Honglei
Chen, Yuting
Hu, Chenpeng
Zhang, Siyue
Shi, Yilei
author_facet Zhang, Honglei
Chen, Yuting
Hu, Chenpeng
Zhang, Siyue
Shi, Yilei
contents As multimodal content continues to expand at a rapid pace, audio retrieval has emerged as a key enabling technology for media search, content organization, and intelligent assistants. However, most existing benchmarks concentrate on semantic matching and fail to capture the fact that real-world queries often demand advanced reasoning abilities, including negation understanding, temporal ordering, concurrent event recognition, and duration discrimination. To address this gap, we introduce ReasonAudio, the first reasoning-intensive benchmark for Text-Audio Retrieval, comprising 1,000 queries and 10,000 composite audio clips across five fundamental reasoning tasks: Negation, Order, Overlap, Duration, and Mix. Despite their intuitive nature for humans and straightforward construction, these tasks pose significant challenges to current models. Our evaluation of ten state-of-the-art models reveals the following findings: All models struggle with reasoning-intensive audio retrieval, performing particularly poorly on Negation and Duration while showing relatively better results on Overlap and Order. Moreover, Multimodal Large Language Model-based embedding models fail to inherit the reasoning capabilities of their backbones through contrastive fine-tuning, suggesting that current training paradigms are insufficient to preserve reasoning capacity in retrieval settings
format Preprint
id arxiv_https___arxiv_org_abs_2605_03361
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle ReasonAudio: A Benchmark for Evaluating Reasoning Beyond Matching in Text-Audio Retrieval
Zhang, Honglei
Chen, Yuting
Hu, Chenpeng
Zhang, Siyue
Shi, Yilei
Artificial Intelligence
As multimodal content continues to expand at a rapid pace, audio retrieval has emerged as a key enabling technology for media search, content organization, and intelligent assistants. However, most existing benchmarks concentrate on semantic matching and fail to capture the fact that real-world queries often demand advanced reasoning abilities, including negation understanding, temporal ordering, concurrent event recognition, and duration discrimination. To address this gap, we introduce ReasonAudio, the first reasoning-intensive benchmark for Text-Audio Retrieval, comprising 1,000 queries and 10,000 composite audio clips across five fundamental reasoning tasks: Negation, Order, Overlap, Duration, and Mix. Despite their intuitive nature for humans and straightforward construction, these tasks pose significant challenges to current models. Our evaluation of ten state-of-the-art models reveals the following findings: All models struggle with reasoning-intensive audio retrieval, performing particularly poorly on Negation and Duration while showing relatively better results on Overlap and Order. Moreover, Multimodal Large Language Model-based embedding models fail to inherit the reasoning capabilities of their backbones through contrastive fine-tuning, suggesting that current training paradigms are insufficient to preserve reasoning capacity in retrieval settings
title ReasonAudio: A Benchmark for Evaluating Reasoning Beyond Matching in Text-Audio Retrieval
topic Artificial Intelligence
url https://arxiv.org/abs/2605.03361