Mieshkov O. Method and system for authentication and driver's functional state evaluation by voice signal

The object of the study is the process of authentication and assessment of a person's functional state when performing professional duties by voice. The purpose of the study is to improve the reliability and speed of the procedures of authentication and assessment of the functional state of the driver in real time by developing new models and methods of analysis of voice signal. Research Methods: Based on the integrated use of analog and digital signal processing theory, mathematical modeling, decision theory, metrology and statistical analysis, computer simulation. Spectral and kepstral analysis were used to determine the main characteristics of the voice signal and to isolate the vocalized areas. Mathematical modeling was used to form localized structures of voice signal in the space of characteristics of "structure-frequency" and to form a personalized voice standard of personality. Decision-making theory was used in the development of the criterion of authentication and determination of the functional state of a person. Metrology theory and statistical analysis were used to evaluate the reliability and effectiveness of the models developed. Computer simulation was used in the software implementation of the methods developed by means of the SciLab 6.0.1 software package. Theoretical results: the first method of authentication and evaluation of the functional state of the person in real time, which consists in the formation of localized structures of a specific person's voice signal by repeated acoustic recording of the voice signal and its processing by the developed methods, followed by a comparison of the formed localized professional responsibilities, which has made it possible to improve the authentication and evaluation of functional procedures the state of the person in the performance of professional duties in real time, in comparison with the existing methods; for the first time, the presentation of human voice signals in the form of localized structures in the combined space of characteristics "structure-frequency" as a combination of points of a given space that correspond to a particular sound of human voice; the method of segmentation of the voice signal has been improved by determining local maxima of the signal spectrum, which takes into account the differences in the spectra of the vocalized regions; The method of framing the vocalized region is improved, characterized in that the frame end point is refined based on the kepstral refinement, followed by the search for the point of transition from the negative to the positive region. The practical results are the development of an algorithm for highlighting the main characteristics of the voice signal, the development of a system of authentication and assessment of the functional state of the driver in real time directly during the performance of professional duties in the form of a combination of hardware and software subsystems, the development of software for making the authentication function and evaluation driver status. Degree of implementation - the results of the dissertation were tested in the form of software-hardware complex for authentication and assessment of functional status at KNP Kherson City Clinical Hospital. OS Luchansky, confirms the high accuracy of the developed methods and means of authentication of the person by voice signal; implemented at the company LLC "Agro-Transit-Invest", which made it possible to improve the safety of operation of vehicles at the enterprise and to improve the working conditions of drivers; in the educational process while teaching the relevant disciplines of the Kherson National Technical University. Sphere (field) of use - medicine.