Employees of the National Research Nizhny Novgorod State University named after N.I. Lobachevsky (UNN) and Tomsk State University of Control Systems and Radioelectronics (TUSUR) have developed a methodology for determining human intoxication by artificial intelligence (AI). This was announced by the author of the study, Valeria Demareva.
According to Demareva, AI can automatically recognize a person's state by their speech. The effectiveness of the methodology does not depend on the language of conversation or the level of background noise.
Methods for determining intoxication by the pitch or amplitude of the voice can become a tool for speech control of call center employees.
Such a control method can improve public safety. Scientists believe that the technology will be in demand in business for monitoring the condition of employees.
The press service of UNN said that an experiment was conducted to evaluate the method. It involved more than 600 people who were in a state of moderate intoxication (1.5 promille of alcohol in the blood).
Volunteers were asked to read tongue twisters, and then the recording of their voices was deciphered using spectral analysis. The data obtained was used to train AI models.
Read more on the topic:
Drivers' intoxication checks are to be tightened in Russia
38% of Russians do not trust AI in hospitals: deputies propose to legislate alternative service