Method and device for replying to robot in outbound call, electronic device and storage medium
By training a large language model and combining audio files and call logs, the problem of outbound call robots being unable to accurately respond to user requests has been solved, enabling more accurate responses in the target domain. This technology is applicable to fields such as insurance, healthcare, technology, and education.
CN116805489BActive Publication Date: 2026-07-03BEIJING WATERDROP TECH GRP CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING WATERDROP TECH GRP CO LTD
- Filing Date
- 2023-06-21
- Publication Date
- 2026-07-03
AI Technical Summary
Technical Problem
Existing outbound call robots are unable to accurately respond to user requests during conversations, resulting in responses that do not match reality.
Method used
By training a pre-set large language model, combining audio files from the target domain with call records from outbound calls, and utilizing the text information converted from the user's request speech, the robot outputs more accurate robot responses.
Benefits of technology
It improves the accuracy of robot responses, enabling better responses to user requests, and is suitable for business needs in fields such as insurance, healthcare, technology, or education.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116805489B_ABST
Abstract
This application discloses a method, apparatus, electronic device, and storage medium for robot responses in outbound calls in some embodiments. The method uses a pre-set large language model trained on audio files from the target domain, making the robot responses obtained through the pre-set large language model more targeted. Furthermore, it inputs existing text information corresponding to all call records in the outbound call, as well as text information converted from the user's request speech, into the pre-set large language model. The pre-set large language model, combined with all call records in the outbound call, outputs robot responses that are more accurate in the target domain in response to the user's request. The method includes: acquiring the user's request speech; converting the request speech into text information; inputting the text information and existing text information into the pre-set large language model to obtain a first output result, and using the first output result as the robot response content.
Need to check novelty before this filing date? Find Prior Art