SpotSound: Enhancing Large Audio-Language Models with Fine-Grained Temporal Grounding

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sun, Luoyi, Zhou, Xiao, Li, Zeqian, Zhang, Ya, Wang, Yanfeng, Xie, Weidi
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915937228161024
author Sun, Luoyi
Zhou, Xiao
Li, Zeqian
Zhang, Ya
Wang, Yanfeng
Xie, Weidi
author_facet Sun, Luoyi
Zhou, Xiao
Li, Zeqian
Zhang, Ya
Wang, Yanfeng
Xie, Weidi
contents Large Audio-Language Models (ALMs) have recently demonstrated remarkable capabilities in holistic audio understanding, yet they remain unreliable for temporal grounding, i.e., the task of pinpointing exactly when an event occurs within long-form audio. This limitation stems from two factors: training data dominated by clip-level supervision lacking precise timestamps, and benchmarks that fail to simulate real-world scenarios where short events are obscured by dense background sounds. In this paper, we introduce SpotSound, an audio language model designed for grounding audio events. SpotSound incorporates a novel training objective, specifically designed to suppress hallucinated timestamps for events absent from the input. Additionally, we present SpotSound-Bench, a challenging temporal grounding benchmark where target events occupy less than ~10\% of each clip, creating a rigorous `needle-in-a-haystack' evaluation. Experiments demonstrate that SpotSound achieves state-of-the-art results on temporal grounding benchmarks while maintaining robust performance across general downstream audio-language tasks. Code, models and benchmark are released on https://loiesun.github.io/spotsound/
format Preprint
id arxiv_https___arxiv_org_abs_2604_13023
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SpotSound: Enhancing Large Audio-Language Models with Fine-Grained Temporal Grounding
Sun, Luoyi
Zhou, Xiao
Li, Zeqian
Zhang, Ya
Wang, Yanfeng
Xie, Weidi
Sound
Multimedia
Large Audio-Language Models (ALMs) have recently demonstrated remarkable capabilities in holistic audio understanding, yet they remain unreliable for temporal grounding, i.e., the task of pinpointing exactly when an event occurs within long-form audio. This limitation stems from two factors: training data dominated by clip-level supervision lacking precise timestamps, and benchmarks that fail to simulate real-world scenarios where short events are obscured by dense background sounds. In this paper, we introduce SpotSound, an audio language model designed for grounding audio events. SpotSound incorporates a novel training objective, specifically designed to suppress hallucinated timestamps for events absent from the input. Additionally, we present SpotSound-Bench, a challenging temporal grounding benchmark where target events occupy less than ~10\% of each clip, creating a rigorous `needle-in-a-haystack' evaluation. Experiments demonstrate that SpotSound achieves state-of-the-art results on temporal grounding benchmarks while maintaining robust performance across general downstream audio-language tasks. Code, models and benchmark are released on https://loiesun.github.io/spotsound/
title SpotSound: Enhancing Large Audio-Language Models with Fine-Grained Temporal Grounding
topic Sound
Multimedia
url https://arxiv.org/abs/2604.13023