Introducing Shizuku-Duplex: A Full-Duplex Audio Synthesis Model for AI Characters

Shizuku-Duplex has been introduced as a full-duplex audio synthesis model built specifically for AI characters to enable more natural human-AI interaction. The model features a novel architecture that manages conversational pacing, asynchronous tool use for background information retrieval, and a compaction mechanism for extended conversations without performance degradation. Open article →