Selvas AI Expands Voice Recognition-Based 'Sexual Violence Victim Investigation Support System' to Police Stations

[Asia Economy Reporter Hyunseok Yoo] AI specialist company Selvas AI announced on the 21st that it will strengthen its public sector capabilities by expanding the application of its AI-based voice recognition statement preparation system to police stations.

Selected as the primary project operator by the National Police Agency last year, Selvas AI successfully commercialized the ‘voice recognition-based sexual violence victim statement preparation system’ in 77 investigation rooms across 59 police stations in Seoul and Gyeonggi areas.

Through this second phase expansion project, the system will be applied to over 90 police stations nationwide, with plans to extend to 257 police stations across the country by next year. Based on the accumulated database (DB), various applied projects such as a question appropriateness assessment tool and integration with the Criminal Justice Information Portal (KICS) will also be pursued.

The 'AI voice recognition statement preparation solution' applied in the ‘AI voice recognition technology-based sexual violence victim investigation support system’ is an AI statement preparation solution that ▲ automatically converts the contents of the testimony statement into text for recording and storage, and converts it into a database (DB). ▲ It recognizes the victim’s spoken testimony by voice and ▲ separates the investigator and victim speakers to document in real time. Additionally, ▲ the system is equipped with an ‘AI Investigation Guide’ that displays related manuals and precedents on the screen according to the victim’s testimony, supporting a more thorough investigation process. It is expected to be distributable not only to police stations but also to various public institutions requiring investigation and inquiry tasks.

Selvas AI is currently advancing system sophistication through the diverse real-world databases (DB) and user feedback collected from the first commercialization service last year. In particular, it is collaborating with Dongguk University, a research partner, to establish a classification system for necessary components by crime type. By specifying step-by-step verification elements such as personal information, disability status, and physical description, it is expected to improve the efficiency of the investigation process.

A Selvas AI official stated, "Following the successful execution of the first commercialization service last year, we have expanded the application to first-class police stations nationwide. Based on this, we will continuously enhance the system to enter various public institutions that require statement preparation." He added, "By adding functions that understand the intentions of investigation subjects and check necessary items during the investigation process, beyond a keyword-centered content system, investigator satisfaction is expected to increase significantly."