From Turn-Taking to Synchronous Dialogue: A Survey of Full-Duplex Spoken Language Models
It addresses the need for more human-like AI interaction by synthesizing fragmented research, but it is incremental as a survey rather than proposing new methods.
This survey tackles the problem of achieving true full-duplex voice communication in AI by reviewing Full-Duplex Spoken Language Models, establishing a taxonomy and evaluation framework, and identifying key challenges like data scarcity and architectural divergence.
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.