AuTAgent: A Reinforcement Learning Framework for Tool-Augmented Audio Reasoning

2026-02-14

AI Summary

This work addresses the challenge that large audio language models (LALMs) struggle to perform precise acoustic measurements in complex audio reasoning tasks, while existing tool-augmented approaches often suffer from information overload or insufficient contextual adaptability. To overcome these limitations, we propose AuTAgent—a reinforcement learning–based agent framework that dynamically decides when and which external audio analysis tools to invoke, guided by sparse feedback and a differential reward mechanism. This approach effectively mitigates redundant tool usage and enhances reasoning accuracy. Experimental results demonstrate that AuTAgent improves the accuracy of both open-source and closed-source base models by 4.20% and 6.20% on the MMAU Test-mini benchmark, and by 9.80% and 8.00% on the MMAR benchmark, respectively, while also exhibiting strong transferability across tasks.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Safety and Robustness
• Multiagent Systems: Multiagent Learning

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Large Audio Language Models (LALMs) excel at perception but struggle with complex reasoning requiring precise acoustic measurements. While external tools can extract fine-grained features like exact tempo or pitch, effective integration remains challenging: naively using all tools causes information overload, while prompt-based selection fails to assess context-dependent utility. To address this, we propose AuTAgent (Audio Tool Agent), a reinforcement learning framework that learns when and which tools to invoke. By employing a sparse-feedback training strategy with a novel Differential Reward mechanism, the agent learns to filter out irrelevant tools and invokes external assistance only when it yields a net performance gain over the base model. Experimental results confirm that AuTAgent complements the representation bottleneck of LALMs by providing verifiable acoustic evidence. It improves accuracy by 4.20% / 6.20% and 9.80% / 8.00% for open-source and closed-source backbones on the MMAU Test-mini and the MMAR benchmarks, respectively. In addition, further experiments demonstrate exceptional transferability. We highlight the complementary role of external tools in augmenting audio model reasoning.

Problem

• Audio Reasoning
• Tool Augmentation
• Large Audio Language Models
• Acoustic Measurement
• Reasoning Bottleneck

Innovation

• Tool-Augmented Reasoning
• Reinforcement Learning
• Audio Language Models
• Differential Reward
• Sparse Feedback