@inproceedings{gu-etal-2026-audio,
title = "Audio-{R}eward-{R}1: Reasoning Reward Model for Unified Audio Understanding and Generation",
author = "Gu, Hao and
Yang, Xiaocheng and
Wu, Jinyang and
Zhao, Hongbo and
Hao, Yunzhe and
Xu, Songcheng and
Bai, Ye and
You, Zhao and
Shang, Shidong",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.144/",
pages = "3327--3348",
ISBN = "979-8-89176-459-0",
abstract = "Multimodal Reward Models are critical for aligning Multimodal Large Language Models with human preferences. However, existing reward models offer limited interpretability and struggle to deliver reliable preference judgments in audio-centric scenarios. To address this gap, we propose \textbf{Audio-Reward-R1}, a reasoning-based reward model that supports both audio understanding and audio generation preference tasks. Audio-Reward-R1 formulates reward modeling as a structured reasoning process, producing dimension-level analyses prior to the final preference judgment. To enable reliable reasoning supervision, we construct a high-quality reasoning-enhanced audio preference dataset with structured reasoning traces aligned with preference annotations. Audio-Reward-R1 is first trained via supervised fine-tuning as a cold-start on this reasoning-enhanced dataset, followed by reinforcement learning with verifiable rewards. Finally, we introduce \textbf{Audio-Reward-Bench}, a human-verified evaluation benchmark for evaluating audio reward models across both audio understanding and audio generation preference tasks. Extensive experimental results demonstrate the effectiveness and reliability of Audio-Reward-R1."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="gu-etal-2026-audio">
<titleInfo>
<title>Audio-Reward-R1: Reasoning Reward Model for Unified Audio Understanding and Generation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hao</namePart>
<namePart type="family">Gu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaocheng</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jinyang</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hongbo</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yunzhe</namePart>
<namePart type="family">Hao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Songcheng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ye</namePart>
<namePart type="family">Bai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhao</namePart>
<namePart type="family">You</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shidong</namePart>
<namePart type="family">Shang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Multimodal Reward Models are critical for aligning Multimodal Large Language Models with human preferences. However, existing reward models offer limited interpretability and struggle to deliver reliable preference judgments in audio-centric scenarios. To address this gap, we propose Audio-Reward-R1, a reasoning-based reward model that supports both audio understanding and audio generation preference tasks. Audio-Reward-R1 formulates reward modeling as a structured reasoning process, producing dimension-level analyses prior to the final preference judgment. To enable reliable reasoning supervision, we construct a high-quality reasoning-enhanced audio preference dataset with structured reasoning traces aligned with preference annotations. Audio-Reward-R1 is first trained via supervised fine-tuning as a cold-start on this reasoning-enhanced dataset, followed by reinforcement learning with verifiable rewards. Finally, we introduce Audio-Reward-Bench, a human-verified evaluation benchmark for evaluating audio reward models across both audio understanding and audio generation preference tasks. Extensive experimental results demonstrate the effectiveness and reliability of Audio-Reward-R1.</abstract>
<identifier type="citekey">gu-etal-2026-audio</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.144/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>3327</start>
<end>3348</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Audio-Reward-R1: Reasoning Reward Model for Unified Audio Understanding and Generation
%A Gu, Hao
%A Yang, Xiaocheng
%A Wu, Jinyang
%A Zhao, Hongbo
%A Hao, Yunzhe
%A Xu, Songcheng
%A Bai, Ye
%A You, Zhao
%A Shang, Shidong
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F gu-etal-2026-audio
%X Multimodal Reward Models are critical for aligning Multimodal Large Language Models with human preferences. However, existing reward models offer limited interpretability and struggle to deliver reliable preference judgments in audio-centric scenarios. To address this gap, we propose Audio-Reward-R1, a reasoning-based reward model that supports both audio understanding and audio generation preference tasks. Audio-Reward-R1 formulates reward modeling as a structured reasoning process, producing dimension-level analyses prior to the final preference judgment. To enable reliable reasoning supervision, we construct a high-quality reasoning-enhanced audio preference dataset with structured reasoning traces aligned with preference annotations. Audio-Reward-R1 is first trained via supervised fine-tuning as a cold-start on this reasoning-enhanced dataset, followed by reinforcement learning with verifiable rewards. Finally, we introduce Audio-Reward-Bench, a human-verified evaluation benchmark for evaluating audio reward models across both audio understanding and audio generation preference tasks. Extensive experimental results demonstrate the effectiveness and reliability of Audio-Reward-R1.
%U https://aclanthology.org/2026.emnlp-main.144/
%P 3327-3348
Markdown (Informal)
[Audio-Reward-R1: Reasoning Reward Model for Unified Audio Understanding and Generation](https://aclanthology.org/2026.emnlp-main.144/) (Gu et al., EMNLP 2026)
ACL
- Hao Gu, Xiaocheng Yang, Jinyang Wu, Hongbo Zhao, Yunzhe Hao, Songcheng Xu, Ye Bai, Zhao You, and Shidong Shang. 2026. Audio-Reward-R1: Reasoning Reward Model for Unified Audio Understanding and Generation. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 3327–3348, Budapest, Hungary. Association for Computational Linguistics.