Research

Paper

AI LLM March 10, 2026

EmoSURA: Towards Accurate Evaluation of Detailed and Long-Context Emotional Speech Captions

Authors

Xin Jing, Andreas Triantafyllopoulos, Jiadong Wang, Shahin Amiriparian, Jun Luo, Björn Schuller

Abstract

Recent advancements in speech captioning models have enabled the generation of rich, fine-grained captions for emotional speech. However, the evaluation of such captions remains a critical bottleneck: traditional N-gram metrics fail to capture semantic nuances, while LLM judges often suffer from reasoning inconsistency and context-collapse when processing long-form descriptions. In this work, we propose EmoSURA, a novel evaluation framework that shifts the paradigm from holistic scoring to atomic verification. EmoSURA decomposes complex captions into Atomic Perceptual Units, which are self-contained statements regarding vocal or emotional attributes, and employs an audio-grounded verification mechanism to validate each unit against the raw speech signal. Furthermore, we address the scarcity of standardized evaluation resources by introducing SURABench, a carefully balanced and stratified benchmark. Our experiments show that EmoSURA achieves a positive correlation with human judgments, offering a more reliable assessment for long-form captions compared to traditional metrics, which demonstrated negative correlations due to their sensitivity to caption length.

Metadata

arXiv ID: 2603.09820
Provider: ARXIV
Primary Category: cs.SD
Published: 2026-03-10
Fetched: 2026-03-11 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.09820v1</id>\n    <title>EmoSURA: Towards Accurate Evaluation of Detailed and Long-Context Emotional Speech Captions</title>\n    <updated>2026-03-10T15:44:48Z</updated>\n    <link href='https://arxiv.org/abs/2603.09820v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.09820v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Recent advancements in speech captioning models have enabled the generation of rich, fine-grained captions for emotional speech. However, the evaluation of such captions remains a critical bottleneck: traditional N-gram metrics fail to capture semantic nuances, while LLM judges often suffer from reasoning inconsistency and context-collapse when processing long-form descriptions. In this work, we propose EmoSURA, a novel evaluation framework that shifts the paradigm from holistic scoring to atomic verification. EmoSURA decomposes complex captions into Atomic Perceptual Units, which are self-contained statements regarding vocal or emotional attributes, and employs an audio-grounded verification mechanism to validate each unit against the raw speech signal. Furthermore, we address the scarcity of standardized evaluation resources by introducing SURABench, a carefully balanced and stratified benchmark. Our experiments show that EmoSURA achieves a positive correlation with human judgments, offering a more reliable assessment for long-form captions compared to traditional metrics, which demonstrated negative correlations due to their sensitivity to caption length.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.SD'/>\n    <published>2026-03-10T15:44:48Z</published>\n    <arxiv:comment>Submitted to Interspeech 2026</arxiv:comment>\n    <arxiv:primary_category term='cs.SD'/>\n    <author>\n      <name>Xin Jing</name>\n    </author>\n    <author>\n      <name>Andreas Triantafyllopoulos</name>\n    </author>\n    <author>\n      <name>Jiadong Wang</name>\n    </author>\n    <author>\n      <name>Shahin Amiriparian</name>\n    </author>\n    <author>\n      <name>Jun Luo</name>\n    </author>\n    <author>\n      <name>Björn Schuller</name>\n    </author>\n  </entry>"
}