arXiv：cs.AI· Jiaheng Dong, Shihong Tan, Gongping Huang, Jean Honorio, Abhirup Ghosh, Ting Dang·· 16 小时前
DuplexAct：前沿全双工语音语言模型的交互能力有多强？
DuplexAct: How Interactive Are Frontier Full-Duplex Spoken Language Models?
AI 导读
DuplexAct 是一个评估端到端全双工语音语言模型（FD-SLM）交互动作能力的基准，定义了涵盖 8 种交互动作、24 种交互条件的三级动作分类体系，并通过即时执行与后续跟进两个维度刻画动作是否成功。该基准包含 4,691 个测试实例，覆盖闲聊、会议、客服和访谈场景，其中 2,531 个来自真实对话录音。
正文
Abstract:End-to-end (E2E) full-duplex spoken language models (FD-SLMs) enable simutanous listening and speaking continously, requiring models to take appropriate interaction actions as conversations unfold. However, existing benchmarks lack a structured action-level formulation, often evaluate action occurrence rather than success, and provide limited coverage of diverse real-world conversations. We introduce DuplexAct, a benchmark for evaluating interaction action taking in E2E FD-SLMs. DuplexAct defines a three-level action taxonomy spanning eight interaction actions and 24 corresponding interaction conditions. We further characterize successful action-taking through immediate execution and subsequent follow-through, and develop action-specific evaluation metrics for both. DuplexAct contains 4,691 test instances spanning casual conversation, meetings, customer service, and interviews, including 2,531 instances derived from real-world conversational recordings. Together, DuplexAct provides a structured and real-world-grounded framework for systematically assessing how E2E FD-SLMs participate in ongoing spoken interaction.
| Comments: | Under review |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.17738 [cs.AI] |
| (or arXiv:2610.17738v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.17738 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Jiaheng Dong [view email]
[v1]
Tue, 13 Oct 2026 05:15:36 UTC (671 KB)
来源：arXiv：cs.AI · arxiv.org