Search academic texts

Information × Registration Number 0219U006010, 0116U005699 , R & D reports Title Modeling of the physical factors influence on biological objects popup.stage_title Head Novikov Oleksandr, Registration Date 27-03-2019 Organization Kherson national technical university popup.description2 The possibility of using a number of technical devices for the recording and primary humn voice signal processing in analogue and digital form has been investigated on the basis of a combined criterion that took into account the frequency range of the microphone and the signal / noise ratio for the received acoustic recording. On the basis of experimental studies, the sampling rate required for the most complete and accurate transformation of the analog signal into digital without loss of information is determined. Its' value, taking into account existing industrial standards, is equal 22050 Hz. The algorithm for allocation of voice signal vocalized areas by means of applied mathematical programs package SciLab 6.0.1 has been developed, which allowed to increase the speed of vocalized areas allocation due to the built-in procedures of direct and inverse Fourier transformations. The algorithm for framing and determining the main frequency of a voice signal by means of applied mathematical programs package SciLab 6.0.1 has been developed, which allowed to increase the speed of determining the basic voice characteristics required for the authentication procedure. An algorithm for the formation of voice signal localized structures in the space of the characteristics "structure-frequency" and the adoption of an authentication solution by means of the applied mathematical package SciLab 6.0.1, which significantly increased the speed of the adoption of authentication solution with a high level of accuracy and reliability. The software-hardware complex for the authentication of the person in the form of a combination of hardware and software subsystems, the main components and blocks of which are the blocks: input sensors and converters, processing of the voice signal, is developed. allocation of basic characteristics of the voice signal; the system of acceptance of the authentication decision, which together with the software and information software implement the developed methods and models. For the first time, the notion of a localized structure of a voice signal as a combination of points of a given space, which corresponds to one or another voice of a particular person's voice, is introduced. The representation of human voice signals in the form of localized structures in the combined space "structure-frequency" characteristics is substantiated. The coordinate axes of this space are the coefficient of the mean square deviation of the signal and the basic frequency of voice. Unlike the standard time or frequency space, this representation allows simultaneously to take into account both the amplitude and frequency characteristics of the voice signal. The method of segmentation of the voice signal is improved by means of determining the local maxima of the signal spectrum, which differs in that it takes into account the differences in the spectra of the vocalized areas, which made it possible to improve the accuracy of the allocation of the limits of the vocalized regions of the voice signal by 6.8% compared with the prototype. The method of framing the vocalized area is improved, which is characterized by the fact that the final point of the frame is specified on the basis of the keplar refinement, followed by the search for the transition point of the signal from the negative to the positive region, which allowed to improve the accuracy of the determination of the basic frequency of the voice signal by 23.2%, and the framemation procedure the signal by 7.0% compared to the existing automatic methods. For the first time, the method of personality authentication is proposed, which consists in the formation of localized structures of a voice signal of a particular person by means of multiple acoustic recording of a voice signal and its processing by developed methods and algorithms with the subsequent comparison of the formed localized structures with the basic ones. For the first time, the notion of a personalized personality voice standard has been introduced, which includes data on the reciprocal placement and the size of localized structures of the voice signal of a particular person in the space of characteristics "structure-frequency". The criterion of personality authentication is proposed, which involves determining the percentage of overlapping area of the formed localized structures of the human voice signal with the basic, which allows to take into account when making an authentication decision of the characteristics of several vocalized sites simultaneously and thus improve the speed and efficiency of the authentication procedure. Product Description popup.authors Мєшков О.Ю. Новіков В.О. popup.nrat_date 2020-04-02 Close