Police to adopt voice phishing analysis system in September

A voice audio data analysis model will be implemented by police from September in voice phishing investigations, the Interior Ministry announced Monday. The model is expected to speed up investigations by rapidly analyzing the voices of perpetrators, the ministry added.

The model, jointly developed by the Interior Ministry and the National Forensic Service in February, is the world's first voice analysis model that specializes in voice phishing data.

The model is also the world's first voice analysis model with "grouping" technology, which analyzes various voices and figures out which criminals belong to the same organization or have ties with each other. This is a crucial part of criminal investigations into cases of voice phishing, as criminals often work in pairs or groups, explained Kim Chul, the head of the data analysis center under the Interior Ministry.

Originally, the investigation agencies had been using an audio analysis model developed in Russia and the UK, but as the model was based on foreign languages, there was a limit to the accuracy of identifying Korean-speaking criminals.

The new model will be distributed through an on-site program that can immediately be used to determine the similarity between collected voice data. The model is 77 percent more accurate than the overseas voice analysis model previously used by the NFS.

During police investigations, this model will compare the voices of criminals that have been already caught and the voices of alleged suspects to identify culprits and uncover further potential crimes.

Earlier, the Interior Ministry and the NFS analyzed 12,323 cases of voice phishing reported to the Financial Supervisory Service from 2015 to March 2023 using the new model. The model identified a total of 5,513 perpetrators of voice phishing crimes, and concluded that 633 of them had participated in 235 different criminal organizations.

Moreover, the Interior Ministry on Tuesday will run a training course on the new voice analysis model at the NFS center in Wonju, Gangwon Province, in a bid to utilize the system more effectively. Officials from 14 organizations, including the main office of the National Police Agency, city and provincial police offices, and prosecutors' offices, will attend this training session to learn how to use the voice analysis technology and interpret the results.