Scientists from Nizhny Novgorod and Tomsk have taught a neural network to determine the level of intoxication in people.

Specially trained artificial intelligence analyzes the voice's sound spectrum—its pitch and amplitude. Scientists claim that the neural network's analysis effectiveness is not reduced by the language spoken or background noise.

According to the head of the cyberpsychology laboratory of the Faculty of Social Sciences of UNN Valeria Demareva, the technology will be in demand in business for assessing the condition of employees whose work involves dynamic speech.

"If someone has communication issues for these reasons, the system will detect this automatically using customer conversation recordings," Demareva explained, adding that this is one way to improve public safety if blood or breath alcohol testing is not available.

The scientists' next goal is to develop algorithms for detecting stress states based on continuous speech. This will help test job applicants' stress tolerance during the hiring process and monitor employee stress levels.