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

arXiv:2605.27741 · cs.CL · Submitted 2026-05-26 · Read on arXiv

cs.CL

Submitted: 2026-05-26

Updated: 2026-10-10

Code: https://github.com/BorrisonXiao/MAPO-release

Terminology

• Benchmark — 2× in this paper · explained in 2 episodes
• Modality-Aware Policy Optimization — 2× in this paper
• Reinforcement Learning — 2× in this paper · explained in 4 episodes
• Fidelity — 1× in this paper · explained in 2 episodes
• GRPO — 1× in this paper · explained in 6 episodes
• Grounding — 1× in this paper · explained in 3 episodes
• Hallucination — 1× in this paper · explained in 4 episodes
• Trace — 1× in this paper · explained in 2 episodes

Sources

• MusicLM: Generating Music From Text
• Audio Flamingo 2: An Audio-Language Model with Long-Audio Understanding and Expert Reasoning Abilities
• Audio Flamingo 3: Advancing Audio Intelligence with Fully Open Large Audio Language Models
• Masur's Divergence for Tori and Kummer Surfaces
• Audio-DeepThinker: Progressive Reasoning-Aware Reinforcement Learning for High-Quality Chain-of-Thought Emergence in Audio Language Models
• Spotlight on Token Perception for Multimodal Reinforcement Learning
• LibriTTS-R: A Restored Multi-Speaker Text-to-Speech Corpus
• MMAU-Pro: A Challenging and Comprehensive Benchmark for Holistic Evaluation of Audio General Intelligence
• MMAR: A Challenging Benchmark for Deep Reasoning in Speech, Audio, Music, and Their Mix
• GPT-4o System Card
• TACOS: Temporally-aligned Audio CaptiOnS for Language-Audio Pretraining
• Direct Preference Optimization: Your Language Model is Secretly a Reward Model
• Omni-R1: Do You Really Need Audio to Fine-Tune Your Audio LLM?
• MMAU: A Massive Multi-Task Audio Understanding and Reasoning Benchmark
• Proximal Policy Optimization Algorithms
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context
• Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities
• MiMo-Audio: Audio Language Models are Few-Shot Learners
• Step-Audio 2 Technical Report

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering