An international group of scientists from Stanford University in the US and the University of Toronto in Canada has developed a technology that can almost guarantee detecting alcohol poisoning in a person from their voice. The study was published in the scientific journal magazine Journal of Alcohol and Drug Research (JSAD).
Experts tested the method on a group of 18 volunteers. The participants were first given a dose of alcohol and then asked to read the nursery rhyme out loud with the voice recording function on the smartphone placed next to them. This procedure was repeated seven times over seven hours.
The scientists then used machine learning to analyze the resulting audio files and identify indicators such as the frequency and pitch of the sound. The algorithms were able to identify volunteers’ alcohol intoxication with 98% accuracy based on voice characteristics.
“I was truly surprised by the accuracy of our model. While we were not pioneers in identifying changes in speech characteristics during alcohol intoxication, I firmly believe that our superior results result from the use of advanced advances in signal processing, acoustic analysis, and machine learning,” said principal investigator Dr. Brian Suffoletto.
According to the scientist, the technology they created is designed to provide timely intervention to prevent road accidents and other accidents. To do this, it is recommended to use ubiquitous devices such as smartphones and smart speakers.
The ultimate goal of the research is to develop a system that people will be willing to use and that will help prevent injuries and save lives, experts said.
Previously Dr. Brian Suffoletto’s team was created A method that allows you to determine whether a person is drunk from his walk using a smartphone.