Automatic labelling system and method of swindle call voice

Abstract

The invention discloses an automatic labelling system of swindle call voice. The system comprises a basic dimension labelling module, a vocal print labelling module and a continuous voice identification labelling module. The output end of the basic dimension labelling module is connected with the input end of the vocal print labelling module and the input end of the continuous voice identificationlabelling module separately. The invention further provides an automatic labelling method of swindle call voice. The method comprises the following specific steps of basic dimension labelling, continuous voice identification and vocal print labelling. According to an intelligent voice technology, input voice data is automatically processed, the voice data is automatically analyzed and identified,the pre-labelling work for labels is conducted, in combination with manual identification, effective labelling management for target data dimension labels is conducted, communication voice data is effectively used, the application and effect of the intelligent voice technology are optimized under the call swindle scene, and meanwhile decryption and encryption transmission of the data in the labelling process is guaranteed.

Description

technical field [0001] The invention relates to a system and method for automatic voice marking of fraudulent calls, belonging to the technical field of fraudulent early warning. Background technique [0002] As a non-contact crime, telephone fraud is highly flexible, changeable, and confrontational. It is difficult to deal with the changes in the form of complex fraudulent methods by relying on a single fixed technical idea. Therefore, it is urgent to use artificial intelligence technology to break through the existing Bottleneck, dig deep into various types of fraud patterns, form a rapid detection and early warning capability of fraud patterns, realize comprehensive discovery of fraud forms, support the upgrade of fraud call management and control models, and improve fraud call management and control capabilities. [0003] At present, domestic telecommunications fraud early warning and prevention mainly include technical routes such as fraud call early warning based on si...

AI Technical Summary

Problems solved by technology

[0005] However, there is currently no standardized data labeling system to guide the labeling work. For example, there are problems such as unclear label dimensions, incomplete coverage, no sensitive decryption and encryption methods, low degree of automatic labeling, and lack of effective verification methods for labeling results.

Images