I am an M.S. student in the Interdisciplinary Program in Artificial Intelligence (IPAI) at Seoul National University, advised by Sungroh Yoon. Previously, I received my B.S. from Seoul National University. My research interests center on speech-language models for natural spoken interaction, with a focus on real-time, human-like agents that can coordinate, interrupt, and continue conversations naturally.
Email: dwsmart32@snu.ac.kr / GitHub / LinkedIn
Consecutive translation can speak in complete phrases, but makes listeners wait.
Simultaneous translation starts sooner, but often breaks speech into short bursts with disruptive pauses.
Can we keep simultaneous translation's low latency while making it as easy to follow as consecutive translation?
What happens when someone nearby tries to hijack the conversation?
How can we build voice assistants that stay robust to third-party interruptions?
Can offline RL generalize to unseen states by reusing the pieces of states it has already seen?
What if an unfamiliar state is not entirely new, but a new composition of familiar anchors and changes?
