Spoken Dialogue

Spoken dialogue research aims to create natural and efficient human-computer interaction through speech, focusing on overcoming limitations of traditional pipeline systems. Current efforts concentrate on developing large language models (LLMs) enhanced with speech processing capabilities, employing techniques like contrastive learning and multimodal fusion to improve understanding of context, speaker identity, and paralinguistic cues. This work is significant for advancing human-computer interaction across various applications, from virtual assistants and customer service to healthcare and autonomous driving, by enabling more robust and nuanced spoken dialogue systems.