Audio cross-modal sentiment analysis method and system based on sentiment database
By constructing an audio cross-modal sentiment analysis method based on a sentiment database, the problems of modal asynchrony and lack of prior knowledge in multimodal sentiment analysis are solved, achieving highly accurate and stable sentiment recognition and adapting to sentiment analysis in complex scenarios.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUAHAN INSTRUMENTS (SHENZHEN) CO LTD
- Filing Date
- 2026-06-02
- Publication Date
- 2026-07-17
AI Technical Summary
Existing multimodal sentiment analysis methods fail to effectively analyze the inherent relationships between audio, video, text semantics, and environmental sensor information, leading to distorted sentiment analysis results and affecting accuracy.
By constructing an audio cross-modal sentiment analysis method based on a sentiment database, this method utilizes multimodal feature extraction, cross-modal time alignment, confidence calculation, prior analysis, and dynamic fusion techniques to generate cross-modal fusion features and output sentiment probability distributions, thus addressing the problems of modal asynchrony, quality fluctuations, and lack of prior knowledge.
It improves the stability, accuracy, and adaptability of the emotion recognition system, enhances the ability to capture subtle emotional changes, and significantly improves the overall accuracy of emotion analysis and its adaptability in complex scenarios.
Smart Images
Figure CN122417092A_ABST