arXiv cs.AI· Daeyong Kwon, Qiyu Wu, Shinobu Kuriya, Junghyun Koo, Shuyang Cui, Zhi Zhong, Wei-Hsiang Liao, Hiromi Wakaki, Yuki Mitsufuji·· 4 小时前AI 评分34
MusTBench：评测与提升音乐大语言模型的时间定位能力
MusTBench: Benchmarking and Advancing Temporal Grounding in Music LLMs
AI 导读
研究者提出 MusTBench 基准，通过五个时间定位问答任务评测大音频-语言模型（LALM）在音乐理解中的时间定位能力，数据经分轨处理与音乐专家验证。实验显示现有 LALM 在精确时间定位上存在困难。团队还提出 MusT 四阶段时间优化方案，涵盖音乐编码器适配、LLM 适配、监督微调和强化学习优化，在不同 LALM 骨干和规模上均带来一致提升。
正文 · 原文
Abstract:Recent Large Audio-Language Models (LALMs) have demonstrated promising abilities in understanding musical content. However, whether their responses are grounded in the correct temporal regions of the audio remains underexplored. This limitation is particularly critical for music understanding, where key information often occurs as temporally localized events, such as instrument entries and rhythmic transitions. To address this gap, we introduce MusTBench, a reliable benchmark with stem-wise data and music-expert validation designed to evaluate temporal grounding in LALMs through five temporally grounded question-answering tasks. To further improve temporal grounding in existing models, we propose MusT, a novel four-stage temporal optimization recipe spanning music encoder adaptation, LLM adaptation, LLM supervised fine-tuning, and RL-based optimization. Experiments on MusTBench show that existing LALMs face challenges in precise temporal grounding, while MusT consistently improves performance across different LALM backbones and scales. These results highlight temporal grounding as an important challenge in current LALMs and provide a benchmark and training recipe that may support future research on temporally grounded music understanding.
| Comments: | Project page: this https URL | Github: this https URL |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Sound (cs.SD) |
| Cite as: | arXiv:2605.29300 [cs.CL] |
| (or arXiv:2605.29300v3 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2605.29300 arXiv-issued DOI via DataCite |
Submission history
From: Daeyong Kwon [view email]
[v1]
Thu, 28 May 2026 03:28:16 UTC (12,011 KB)
[v2]
Tue, 1 Sep 2026 15:58:50 UTC (12,003 KB)
[v3]
Wed, 4 Nov 2026 13:56:29 UTC (9,735 KB)
来源：arXiv cs.AI · arxiv.org