SEOUL-- The government on Wednesday unveiled a new artificial intelligence-based voice analysis model aimed at combating voice phishing scams.

The program was jointly developed by the Ministry of the Interior and Safety's integrated data analysis center and the National Forensic Service, and will be deployed in investigations later this month, the ministry said during a news conference in the central city of Sejong.

The AI model compares the voices of suspected scammers with existing voice data from known criminals.

The National Forensic Service has utilized voice analysis models developed in Russia and Britain, but the models' accuracy in identifying Korean-speaking fraudsters was low, at around 30 percent, due to language differences.

The new program utilizes more than 1 million Korean and foreign language voices extracted from over 6,000 scammers and ordinary people.

By leveraging deep learning technology, the new model has improved the accuracy of identification by 77 percent compared to existing foreign-made models, the ministry said.

Furthermore, the AI program can classify criminals impersonating different roles, allowing investigators to better identify individuals belonging to the same organizations.

The voices of swindlers analyzed by the new model will be released on the website of the Financial Supervisory Service.

The ministry said it plans to promote the program overseas through international events later this year.

Source: Yonhap News Agency