Real-TurnTurk Multimodal Dataset for Turkish Turn-Taking Prediction
August 21, 2026
Real-TurnTurk is a multimodal Turkish conversational corpus designed to improve turn-taking prediction in synchronous dialog systems. It includes synchronized video, per-speaker audio for overlapping speech, and time-aligned transcriptions.
HOW THIS AFFECTS YOU
●
researcherYou can use this to train models on the nuances of naturalistic, unscripted dyadic interactions in Turkish.
●
designerThis can help improve the conversational flow and naturalness of voice-based UX in the Turkish market.