Audio Reasoning

Follow this topic Ask about this topic

Momentum

12 papers in the last four weeks, up 200% on the four weeks before. 0.1% of all new papers.

Jul 13Week of Sep 28

Latest papers 41

All topics

Escape the Language Prior: Mitigating Late-Stage Modality Collapse in Audio Reasoning via Modality-Aware Policy Optimization
May 26, 2026Cihan Xiao, Yiwen Shao, Chenxing Li +5RL for Language Model Reasoning Audio Reasoning
Johns Hopkins University · Tencent Hunyuan
Learning When to Think While Listening in Large Audio-Language Models
May 26, 2026Zhiyuan Song, Weici Zhao, Yang Xiao +3Audio QA RL for Language Model Reasoning
University of Pennsylvania
When Vision Speaks for Sound
May 13, 2026Xiaofei Wen, Wenjie Jacky Mo, Xingyu Fu +6Audio-Language Model Evaluation Audio-Visual Understanding
dUniversity of California, Davis · pPrinceton University · wUniversity of Wisconsin–Madison +1
Afrispeech Semantics: Evaluating Audio Semantic Reasoning in Spoken Language Models Across Domains and Accents
May 11, 2026Chibuzor Okocha, Christan GrantAudio-Language Model Evaluation Spoken Language Understanding
University of Florida
ReasonAudio: A Benchmark for Evaluating Reasoning Beyond Matching in Text-Audio Retrieval
May 5, 2026Honglei Zhang, Yuting Chen, Chenpeng Hu +2Audio Reasoning Audio Understanding
School of Software, Nanjing University Nanjing, China · School of Software, Northwestern Polytechnical University Xi’an, China · College of Computing and Data Science, Nanyang Technological University Singapore, Singapore
AUDITA: A New Dataset to Audit Humans vs. AI Skill at Audio QA
Apr 23, 2026Tasnim Kabir, Dmytro Kurdydyk, Aadi Palnitkar +3Audio QA Audio Reasoning
University of Maryland · Davidson College · Columbia University +1
Audio-DeepThinker: Progressive Reasoning-Aware Reinforcement Learning for High-Quality Chain-of-Thought Emergence in Audio Language Models
Apr 20, 2026Xiang He, Chenxing Li, Jinting Wang +5RL for Language Model Reasoning Audio Reasoning
Tencent AI Lab, Beijing, China · The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China
Listen, Pause, and Reason: Toward Perception-Grounded Hybrid Reasoning for Audio Understanding
Apr 16, 2026Jieyi Wang, Yazhe Niu, Dexuan Xu +1Audio QA Audio Reasoning
Shanghai AI Laboratory · Peking University · CUHK MMLab +1
PARSA-Bench: A Comprehensive Persian Audio-Language Model Benchmark
Mar 15, 2026Mohammad Javad Ranjbar Kalahroodi, Mohammad Amini, Parmis Bathayan +2Audio-Language Model Evaluation Audio Reasoning
School of Electrical and Computer Engineering, University of Tehran, Iran · Institute for Research in Fundamental Sciences (IPM), Tehran, Iran
When Scaling Fails: Mitigating Audio Perception Decay of LALMs via Multi-Step Perception-Aware Reasoning
Feb 28, 2026Ruixiang Mao, Xiangnan Ma, Dan Chen +12Audio Reasoning Audio Understanding
Northeastern University,China · NiuTrans Research.
TAC: Timestamped Audio Captioning
Feb 17, 2026Sonal Kumar, Prem Seetharaman, Ke Chen +8Audio Captioning Temporal Audio Grounding
University of Maryland, College Park, USA · Adobe Research, USA · OpenAI, USA (work done while at Adobe).

May 26, 2026, cs.CLJ/K move · Enter open · S save

Escape the Language Prior: Mitigating Late-Stage Modality Collapse in Audio Reasoning via Modality-Aware Policy Optimization

Cihan Xiao, Yiwen Shao, Chenxing Li, Xiang He+4

Johns Hopkins University · Tencent Hunyuan

RL for Language Model Reasoning Audio Reasoning Audio-Language Models Multimodal Reasoning

Open paper AskPDF arXiv

Audio and omni-modal large language models exhibit impressive cross-modal reasoning capabilities. However, applying standard reinforcement learning post-training algorithms to these models exposes a critical structural vulnerability: methods like GRPO apply uniform policy gradients across all tokens, ignoring their unequal dependence on the non-text source modality. This exacerbates late-stage modality collapse during extended chain-of-thought generation, where models progressively abandon the primary source signal in favor of compressed textual priors, leading to confident but ungrounded hallucinations. To address this, we introduce Modality-Aware Policy Optimization (MAPO), a novel dual-branch reinforcement learning framework. First, MAPO dynamically concentrates the policy gradient on modality-critical tokens using a modality relevance mask, which is derived from the cross-modal differential entropy between an audio-ablated reference and the multimodal policy. Second, it integrates an auxiliary attention loss branch that applies a targeted, temporally scaled penalty to the model's internal attention distributions. This ensures the model actively sustains cross-modal grounding deep into the reasoning trace. Evaluations on complex audio reasoning benchmarks demonstrate that MAPO substantially improves long-horizon reasoning fidelity and multimodal instruction following, achieving highly competitive performance and setting new state-of-the-art results on several key benchmarks among open-weight models. By relying strictly on native statistical signals rather than domain-specific inductive biases, MAPO offers a promising foundation for mitigating epistemic collapse across diverse multimodal systems.

Momentum

Related topics

Latest papers 41