一种面向虚拟现实的多源融合情感支持对话方法
By collecting and fusing multi-source features from speech and eye-tracking information, high-quality emotional dialogue responses are generated, solving the problem of emotional dialogues not fitting real-world scenarios in existing technologies and improving the user interaction experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SUN YAT SEN UNIV
- Filing Date
- 2023-11-21
- Publication Date
- 2026-07-17
AI Technical Summary
Existing emotional dialogue generation technologies produce low-quality emotional dialogues that do not closely resemble real-life dialogue scenarios and lack multimodal information fusion, resulting in a poor user interaction experience.
The system collects users' voice and eye movement information, extracts text, audio, and eye movement feature vectors, performs feature fusion through a multi-subspace shared private representation model, generates emotion representation vectors, and uses a GPT-3 decoder to generate dialogue responses that conform to contextual semantics and emotional color.
It improves the quality of emotional dialogue, making it more realistic and enhancing the user's interactive experience. Through multi-source interaction methods, including visual and auditory elements, it enhances the user's immersion and emotional feedback.
Smart Images
Figure CN117370534B_ABST