In Russia, specialists from the Artificial Intelligence Institute AIRI and the Moscow Technical University of Communications and Informatics (MTUCI) have developed an AI-based system to combat telephone scammers using synthetic voices.
The new technology is capable of recognizing voice forgery with high accuracy, which is created using modern text-to-speech technologies. This system helps to distinguish a genuine voice from a synthetic one, taking into account the context and characteristics of voice data, and adapts to new types of threats.
The development can be implemented in various security systems to protect against attacks using voice spoofing. It can also be used to improve the reliability of voice authentication. The Russian model has already demonstrated its effectiveness, surpassing similar developments by scientists from South Korea and France in terms of performance. The system was among the top ten solutions at the international Asvspoof 2024 Challenge.
Read more on the topic:
AI fraudsters extort money using fabricated voices of SUSU management
Yandex neural networks are seven times faster at finding fraudsters
Hacking Telegram with AI - how fraudsters in Russia use your voice and passport