Customer call emotion real-time analysis method and device based on deep learning, storage medium and computer equipment
By integrating multimodal information from acoustics, speech rate, and text semantics through deep learning, and utilizing cross-modal attention mechanisms and an end-to-end architecture, the accuracy and real-time performance of emotion analysis in customer service calls have been addressed. This enables instant emotion feedback and strategy adjustment, improving the precision and timeliness of service interactions.
Patent Information
- Application Number
- CN202610016588.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-07
- Publication Date
- 2026-05-26
AI Technical Summary
Existing customer service call sentiment analysis technologies rely on single-modal analysis, which makes it difficult to accurately identify customers' true emotions in complex contexts. Furthermore, they lack real-time processing capabilities, resulting in high analysis latency and an inability to adjust communication strategies in a timely manner.
By integrating multimodal information from acoustics, speech rate, and text semantics through deep learning, and dynamically fusing features using a cross-modal attention mechanism, combined with an end-to-end real-time processing architecture, we can acquire and analyze multidimensional emotional states in customer service calls in real time.
It enables real-time output of customer emotional state during calls, allowing customer service personnel to adjust communication strategies promptly, significantly improving the accuracy and timeliness of service interactions.
Smart Images

Figure CN122090881A_ABST
