从轮流对话到同步对话：全双工口语语言模型综述
From Turn-Taking to Synchronous Dialogue: A Survey of Full-Duplex Spoken Language Models
AI总结：
本文综述了LLM时代的全双工口语语言模型，提出工程化同步与学习式同步的分类法，并统一评估框架，指出同步数据稀缺、架构分歧和评估缺口等核心挑战，为人机通信发展提供路线图。
摘要：
真正的全双工（TFD）语音通信——能够在自然轮流对话、重叠语音和打断的情况下同时听与说——代表着迈向类人AI交互的一个关键里程碑。本综述全面回顾了大语言模型（LLM）时代的全双工口语语言模型（FD-SLMs）。我们建立了一种分类法，区分工程化同步（模块化架构）与学习式同步（端到端架构），并将碎片化的评估方法统一到一个涵盖时间动态、行为仲裁、语义连贯性和声学性能的框架中。通过对主流FD-SLMs的比较分析，我们识别出根本性挑战：同步数据稀缺、架构分歧和评估缺口，从而为推进人机通信提供了路线图。
英文摘要：
True Full-Duplex (TFD) voice communication--enabling simultaneous listening and speaking with natural turn-taking, overlapping speech, and interruptions--represents a critical milestone toward human-like AI interaction. This survey comprehensively reviews Full-Duplex Spoken Language Models (FD-SLMs) in the LLM era. We establish a taxonomy distinguishing Engineered Synchronization (modular architectures) from Learned Synchronization (end-to-end architectures), and unify fragmented evaluation approaches into a framework encompassing Temporal Dynamics, Behavioral Arbitration, Semantic Coherence, and Acoustic Performance. Through comparative analysis of mainstream FD-SLMs, we identify fundamental challenges: synchronous data scarcity, architectural divergence, and evaluation gaps, providing a roadmap for advancing human-AI communication.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
