cff-version: 1.2.0 message: "If you use this software, evaluation benchmark, or full-duplex voice agent workbench in your research, please cite it using the metadata below." type: software title: "Proactive Turn-Taking at Transition Relevance Places for Real-Time Conversational Agents: Benchmark & Full-Duplex Voice Agent Workbench" version: "1.0.0" date-released: "2026-10-18" license: Apache-2.0 repository-code: "https://github.com/yixinzhangagent/proactive-turn-taking-benchmark" url: "https://yixinzhangagent.github.io" authors: - family-names: "Zhang" given-names: "Yixin" email: "zyxcambridge@gmail.com" affiliation: "Yuansiwujie" orcid: "https://orcid.org/0009-0004-7641-4450" keywords: - "proactive turn-taking" - "transition relevance place" - "full-duplex voice agent" - "conversational AI" - "deliberate silence" - "multiplicative utility model" - "evaluation benchmark" abstract: >- Reproducible evaluation benchmark (200 multi-turn task-oriented dialogue trajectories, 480 intra-turn thinking pauses, and 600 Transition Relevance Places across four domains) and pure-NumPy 24/7 full-duplex voice agent runtime workbench implementing the Three-Factor Multiplicative Utility Model U(t) = A(t) * S(t) * N(t) and the Gricean Three-Question Validity Filter for proactive turn-taking at Transition Relevance Places (TRPs). preferred-citation: type: article title: "Proactive Turn-Taking at Transition Relevance Places for Real-Time Conversational Agents" authors: - family-names: "Zhang" given-names: "Yixin" email: "zyxcambridge@gmail.com" affiliation: "Yuansiwujie" orcid: "https://orcid.org/0009-0004-7641-4450" year: 2026 journal: "arXiv preprint" url: "https://github.com/yixinzhangagent/proactive-turn-taking-benchmark"