AudioToolAgent: An Agentic Framework for Audio-Language Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914328183046144 |
|---|---|
| author | Wijngaard, Gijs Formisano, Elia Dumontier, Michel Jitsev, Jenia |
| author_facet | Wijngaard, Gijs Formisano, Elia Dumontier, Michel Jitsev, Jenia |
| contents | Large Audio-Language Models (LALMs) perform well on audio understanding tasks but lack multistep reasoning and tool-calling found in recent Large Language Models (LLMs). This paper presents AudioToolAgent, a framework that coordinates audio-language models as tools via a central LLM agent that accesses tool adapters for audio question answering and speech-to-text. The agent reasons about which tools to invoke, how to formulate follow-up queries, and how to arbitrate conflicting tool outputs, without accessing the audio. Experiments with MMAU, MMAR, and MMAU-Pro show state-of-the-art accuracy: up to 77.50% in MMAU, 77.00% in MMAR, and 61.90% in MMAU-Pro. Shapley-based analysis identifies effective agent-tool combinations. The code and reproduction materials are available at https://github.com/GLJS/AudioToolAgent. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_02995 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | AudioToolAgent: An Agentic Framework for Audio-Language Models Wijngaard, Gijs Formisano, Elia Dumontier, Michel Jitsev, Jenia Sound Large Audio-Language Models (LALMs) perform well on audio understanding tasks but lack multistep reasoning and tool-calling found in recent Large Language Models (LLMs). This paper presents AudioToolAgent, a framework that coordinates audio-language models as tools via a central LLM agent that accesses tool adapters for audio question answering and speech-to-text. The agent reasons about which tools to invoke, how to formulate follow-up queries, and how to arbitrate conflicting tool outputs, without accessing the audio. Experiments with MMAU, MMAR, and MMAU-Pro show state-of-the-art accuracy: up to 77.50% in MMAU, 77.00% in MMAR, and 61.90% in MMAU-Pro. Shapley-based analysis identifies effective agent-tool combinations. The code and reproduction materials are available at https://github.com/GLJS/AudioToolAgent. |
| title | AudioToolAgent: An Agentic Framework for Audio-Language Models |
| topic | Sound |
| url | https://arxiv.org/abs/2510.02995 |