当扩展失败时：通过多步感知感知推理缓解大型音频语言模型的音频感知衰减

Computer Science > Sound arXiv:2603.02266 (cs) [Submitted on 28 Feb 2026 (v1), last revised 21 Sep 2026 (this version, v2)] Title:When Scaling Fails: Mitigating Audio Perception Decay of LALMs via Multi-Step Perception-Aware Reasoning Authors:Ruixiang Mao, Xiangnan Ma, Dan Chen, Ziming Zhu, Yuan Ge, Aokai Hao, Haishu Zhao, Yifu Huo, Qing Yang, Kaiyan Chang, Xiaoqian Liu, Chenglong Wang, Qiaozhi He, Tong Xiao, Jingbo Zhu View a PDF of the paper titled When Scaling Fails: Mitigating Audio Perception Decay of LALMs via Multi-Step Perception-Aware Reasoning, by Ruixiang Mao and 14 other authors View PDF HTML (experimental) Abstract:Test-Time Scaling has shown notable efficacy in addressing complex problems through scaling inference compute. However, within Large Audio-Language Models (LALMs), an unintuitive phenomenon exists: post-training models for structured reasoning trajectories results in marginal or even negative gains compared to post-training for direct answering. To investigate it, we introduce CAFE, an evaluation framework designed to precisely quantify audio reasoning errors. Evaluation results reveal LALMs struggle with perception during reasoning and encounter a critical bottleneck: reasoning performance suffers from audio perception decay as reasoning length extends. To address it, we propose MPAR$^2$, a paradigm that encourages dynamic perceptual reasoning and decomposes complex questions into perception-rich sub-problems. Leveraging reinforcement learning, MPAR$^2$ improves perception performance on CAFE from 31.74% to 63.51% and effectively mitigates perception decay, concurrently enhancing reasoning capabilities to achieve a significant 74.59% accuracy on the MMAU benchmark. Further analysis demonstrates that MPAR$^2$ reinforces LALMs to attend to audio input and dynamically adapts reasoning budget to match task complexity. Comments: Subjects: Sound (cs.SD); Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS) Cite as: arXiv:2603.02266 [cs.SD] (or arXiv:2603.02266v2 [cs.SD] for this version) https://doi.org/10.48550/arXiv.2603.02266 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Ruixiang Mao [view email] [v1] Sat, 28 Feb 2026 12:54:51 UTC (1,704 KB) [v2] Mon, 21 Sep 2026 15:40:24 UTC (1,704 KB) Full-text links: Access Paper: View a PDF of the paper titled When Scaling Fails: Mitigating Audio Perception Decay of LALMs via Multi-Step Perception-Aware Reasoning, by Ruixiang Mao and 14 other authors View PDF HTML (experimental) TeX Source view license Current browse context: cs.SD new | recent | 2026-03 Change to browse by: cs cs.AI eess eess.AS References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)