dialogue-dataset

DuplexDrama is the first synthesized spoken dialogue dataset that covers personas, scenarios, full-duplex behaviors, expressive speech, and sound events, designed to advance full-duplex dialogue model research with over 2,000 hours of audio data.

This paper introduces a French OSCE dialogue dataset of 240 interactions and a controllable LLM-based pipeline for generating synthetic OSCE dialogues, enabling realistic virtual patient simulations for medical training with automatic feedback.