講演情報
[3O1-GS-10v-03]Voice Agent Technology Based on High-Quality Conversational Dataset
〇Qingqing Zhang1 (1. Beijing Magic Data Technology Co., Ltd.)
キーワード：
AI、Voice Agent、High-Quality Conversational Dataset、Human-Robot Interaction
Recent advances in generative artificial intelligence have significantly expanded the role of speech-based interaction in robotic systems and embodied agents. However, existing speech and dialogue models often struggle to achieve natural multi-turn interaction, emotional awareness, contextual coherence, and stable agent identity in real-world environments. This limitation is largely attributed to the lack of high-quality, multimodal conversational data that accurately reflects human communication behaviors.
Experimental analysis demonstrates that training with natural, multimodal conversational data substantially improves response latency, conversational coherence, emotional expressiveness, and robustness in complex interaction scenarios. Our findings highlight the critical role of data-centric design in advancing next-generation voice agents for robotic systems.
Experimental analysis demonstrates that training with natural, multimodal conversational data substantially improves response latency, conversational coherence, emotional expressiveness, and robustness in complex interaction scenarios. Our findings highlight the critical role of data-centric design in advancing next-generation voice agents for robotic systems.
コメント
コメントの閲覧・投稿にはログインが必要です。ログイン