The Study of an Impact of Alcohol Intoxication on Voice Parameters in Biometric Authentication Systems

Changing person voice characteristics under the influence of alcohol intoxication affects the effectiveness of the voice identification procedures. Therefore, determining the level of alcohol in the blood remotely allows not only to identify the fact that the subject is in an altered psychophysiological state but also makes it possible to adjust the settings of the decision-making algorithm to increase the reliability of voice authentication. The purpose of the work is to analyze existing scientific works and results in the field of assessing the level of alcohol intoxication by voice and the analysis of data sets that make it possible to train artificial intelligence models to identify the fact of intoxication, as well as the analysis of factors influencing the psychophysiological state of the speaker on the parameters of his speech. The study includes a comparative analysis of scientific works from databases such as Scopus, Web of Science, and articles included in the list of Higher Attestation Commissions and takes into account factors such as gender, age, stages of alcohol intoxication, recording quality, and ambient noise level. Based on an analysis of scientific publications, the study identifies random forest as one of the most effective machine learning methods, demonstrating an accuracy of 95.3% on proprietary speech datasets and 80% on the widely used alcohol language corpus.