# SARI: Structured Audio Reasoning via Curriculum-Guided Reinforcement Learning arXiv: https://arxiv.org/abs/2504.15900 ยท HTML: https://mlsanity.com/paper/2504.15900/ ## Key facts - Cited by 34 papers in the map (data through 14 Oct 2026). - Placed in Audio-language model reasoning, one of 17 directions in LLM reasoning. - Builds on 26 of its 32 references that are in the map. - First posted on arXiv 22 Apr 2025. ## Placed in - LLM reasoning / [Audio-language model reasoning](https://mlsanity.com/lr/audio-language-model-reasoning/) ## Builds on (most cited first, 26 of 26) - [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022) - [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024) - [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025) - [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024) - [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://mlsanity.com/paper/2503.14476/) (2025) - [Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning](https://mlsanity.com/paper/2503.09516/) (2025) - [Kimi k1.5: Scaling Reinforcement Learning with LLMs](https://mlsanity.com/paper/2501.12599/) (2025) - [Measuring and Narrowing the Compositionality Gap in Language Models](https://mlsanity.com/paper/2210.03350/) (2022) - [Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debate](https://mlsanity.com/paper/2305.19118/) (2023) - [Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Models](https://mlsanity.com/paper/2503.06749/) (2025) - [MELD: A Multimodal Multi-Party Dataset for Emotion Recognition in Conversations](https://mlsanity.com/paper/1810.02508/) (2018) - [Qwen2-Audio Technical Report](https://mlsanity.com/paper/2407.10759/) (2024) - [SALMONN: Towards Generic Hearing Abilities for Large Language Models](https://mlsanity.com/paper/2310.13289/) (2023) - [Video-R1: Reinforcing Video Reasoning in MLLMs](https://mlsanity.com/paper/2503.21776/) (2025) - [Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Models](https://mlsanity.com/paper/2503.09567/) (2025) - [R1-Onevision: Advancing Generalized Multimodal Reasoning through Cross-Modal Formalization](https://mlsanity.com/paper/2503.10615/) (2025) - [MMAU: A Massive Multi-Task Audio Understanding and Reasoning Benchmark](https://mlsanity.com/paper/2410.19168/) (2024) - [Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement Learning](https://mlsanity.com/paper/2502.14768/) (2025) - [LMM-R1: Empowering 3B LMMs with Strong Reasoning Abilities Through Two-Stage Rule-Based RL](https://mlsanity.com/paper/2503.07536/) (2025) - [VoiceBench: Benchmarking LLM-Based Voice Assistants](https://mlsanity.com/paper/2410.17196/) (2024) - [Audio Flamingo: A Novel Audio Language Model with Few-Shot Learning and Dialogue Abilities](https://mlsanity.com/paper/2402.01831/) (2024) - [Mustango: Toward Controllable Text-to-Music Generation](https://mlsanity.com/paper/2311.08355/) (2023) - [Audio-Reasoner: Improving Reasoning Capability in Large Audio Language Models](https://mlsanity.com/paper/2503.02318/) (2025) - [Reinforcement Learning Outperforms Supervised Fine-Tuning: A Case Study on Audio Question Answering](https://mlsanity.com/paper/2503.11197/) (2025) - [Audio-CoT: Exploring Chain-of-Thought Reasoning in Large Audio Language Model](https://mlsanity.com/paper/2501.07246/) (2025) - [ThinkPatterns-21k: A Systematic Study on the Impact of Thinking Patterns in LLMs](https://mlsanity.com/paper/2503.12918/) (2025) ## Built on by (most cited first, 30 of 34) - [The Landscape of Agentic Reinforcement Learning for LLMs: A Survey](https://mlsanity.com/paper/2509.02547/) (2025) - [MMAR: A Challenging Benchmark for Deep Reasoning in Speech, Audio, Music, and Their Mix](https://mlsanity.com/paper/2505.13032/) (2025) - [Perception, Reason, Think, and Plan: A Survey on Large Multimodal Reasoning Models](https://mlsanity.com/paper/2505.04921/) (2025) - [LongVT: Incentivizing "Thinking with Long Videos" via Native Tool Calling](https://mlsanity.com/paper/2511.20785/) (2025) - [Multimodal Spatial Reasoning in the Large Model Era: A Survey and Benchmarks](https://mlsanity.com/paper/2510.25760/) (2025) - [STITCH: Simultaneous Thinking and Talking with Chunked Reasoning for Spoken Language Models](https://mlsanity.com/paper/2507.15375/) (2025) - [Audio-Thinker: Guiding Audio Language Model When and How to Think via Reinforcement Learning](https://mlsanity.com/paper/2508.08039/) (2025) - [Reinforcement Fine-Tuning Powers Reasoning Capability of Multimodal Large Language Models](https://mlsanity.com/paper/2505.18536/) (2025) - [BOTS: A Unified Framework for Bayesian Online Task Selection in LLM Reinforcement Finetuning](https://mlsanity.com/paper/2510.26374/) (2025) - [Mind-Paced Speaking: A Dual-Brain Approach to Real-Time Reasoning in Spoken Language Models](https://mlsanity.com/paper/2510.09592/) (2025) - [Audio Flamingo Sound-CoT Technical Report: Improving Chain-of-Thought Reasoning in Sound Understanding](https://mlsanity.com/paper/2508.11818/) (2025) - [Echo: Towards Advanced Audio Comprehension via Audio-Interleaved Reasoning](https://mlsanity.com/paper/2602.11909/) (2026) - [Omni-CLST: Error-aware Curriculum Learning with guided Selective chain-of-Thought for audio question answering](https://mlsanity.com/paper/2509.12275/) (2025) - [Teaching Audio Models to Reason: A Unified Framework for Source- and Layer-wise Distillation](https://mlsanity.com/paper/2509.18579/) (2025) - [Learning When to Think While Listening in Large Audio-Language Models](https://mlsanity.com/paper/2605.27190/) (2026) - [Kalman Meets Curriculum: Efficient Dynamic Prompt Selection for Adaptive RL Finetuning](https://mlsanity.com/paper/2607.27610/) (2026) - [EvoAudio: Recursive Self-Improvement for Audio Understanding](https://mlsanity.com/paper/2609.27389/) (2026) - [Enhancing Speech Large Language Models through Reinforced Behavior Alignment](https://mlsanity.com/paper/2509.03526/) (2025) - [SightSound-R1: Cross-Modal Reasoning Distillation from Vision to Audio Language Models](https://mlsanity.com/paper/2509.15661/) (2025) - [Rebellion: Noise-Robust Reasoning Training for Audio Reasoning Models](https://mlsanity.com/paper/2511.09682/) (2025) - [MathMixup: Boosting LLM Mathematical Reasoning with Difficulty-Controllable Data Synthesis and Curriculum Learning](https://mlsanity.com/paper/2601.17006/) (2026) - [ADORA: Training Reasoning Models with Dynamic Advantage Estimation on Reinforcement Learning](https://mlsanity.com/paper/2602.10019/) (2026) - [When Scaling Fails: Mitigating Audio Perception Decay of LALMs via Multi-Step Perception-Aware Reasoning](https://mlsanity.com/paper/2603.02266/) (2026) - [AVRT: Audio-Visual Reasoning Transfer through Single-Modality Teachers](https://mlsanity.com/paper/2604.16617/) (2026) - [Disentangling Reasoning in Large Audio-Language Models for Ambiguous Emotion Prediction](https://mlsanity.com/paper/2603.08230/) (2026) - [AudioDER: A Deduplication-Enhanced Reasoning Dataset for Post-Training Large Audio-Language Models](https://mlsanity.com/paper/2606.14591/) (2026) - [CMA-OT: Hierarchical Expert Supervision for Dance-to-Music Generation](https://mlsanity.com/paper/2609.13118/) (2026) - [Listen Then Reason: Perception-Grounded Test-Time Reinforcement Learning for Large Audio-Language Models](https://mlsanity.com/paper/2609.23589/) (2026) - [Spoken Language Models that Think Aloud](https://mlsanity.com/paper/2609.26488/) (2026) - [Mizar: A 159M-Parameter Audio-Language Model for Audio Understanding](https://mlsanity.com/paper/2609.28344/) (2026) ## Abstract (arXiv) Recent work shows that reinforcement learning(RL) can markedly sharpen the reasoning ability of large language models (LLMs) by prompting them to "think before answering." Yet whether and how these gains transfer to audio-language reasoning remains largely unexplored. We extend the Group-Relative Policy Optimization (GRPO) framework from DeepSeek-R1 to a Large Audio-Language Model (LALM), and construct a 32k sample multiple-choice corpus. Using a two-stage regimen supervised fine-tuning on structured and unstructured chains-of-thought, followed by curriculum-guided GRPO, we systematically compare implicit vs. explicit, and structured vs. free form reasoning under identical architectures. Our structured audio reasoning model, SARI (Structured Audio Reasoning via Curriculum-Guided Reinforcement Learning), achieves a 16.35% improvement in average accuracy over the base model Qwen2-Audio-7B-Instruct. Furthermore, the variant built upon Qwen2.5-Omni reaches state-of-the-art performance of 67.08% on the MMAU test-mini benchmark. Ablation experiments show that on the base model we use: (i) SFT warm-up is important for stable RL training, (ii) structured chains yield more robust generalization than unstructured ones, and (iii) easy-to-hard curricula accelerate convergence and improve final performance. These findings demonstrate that explicit, structured reasoning and curriculum learning substantially enhances audio-language understanding.