Article 7116

Title of the article

AN ALGORITHM FOR CONSTRUCTING A SPEECH SOUND DEFRAGMENTER FOR VOICE RECOGNITION TAKING INTO ACCOUNT BIOMETRIC FEATURES OF SPEAKERS 

Authors

Boykov Il'ya Vladimirovich, Doctor of physical and mathematical sciences, professor, head of sub-department of higher and applied mathematics, Penza State University (40 Krasnaya street, Penza, Russia), boikov@pnzgu.ru
Kalashnikov Dmitriy Mikhaylovich, Postgraduate student, Penza State University (40 Krasnaya street, Penza, Russia), kalashnikovdm.penza@gmail.com

Index UDK

004; 519.7; 519.6; 519.66; 612.087.1

Abstract

Background. Recently the question of ensuring information security is particularly acute. Voice identification of personality hasn't become current so far because of a number of unresolved problems. One of the major problems is reliability of authentication. Now the probability of an error of recognition of speaker’s voice is rather high. There is a need for algorithms that more accurately identify biometric parameters of speakers by voice signals. The second problem is unstable operation of the equipment in conditions of noise. The third problem is made by a variety of manifestations of one person’s voice: the voice is capable to change depending on a state of health, age, mood etc. The present work offers methods and algorithms directed to solve these problems.
Materials and methods. The authors used numerical methods of continuous and discrete information processing, methods of harmonious analysis, spectral methods, methods of mathematical statistics and temporary ranks. The continual and discrete model of speech processing, in combination with the narrow-band filter, allowing to determine the average length of sound, was taken as the basis for creation of a fragmentator. The researchers used linear predata processing of voice signals for specification of the period of the main tone.
Results. The work offers the method of determination of speaker’s identity by the results of the analysis of speech fragments. The new method of speech fragmentation in general and separate phrases is offered. Introduction of this method of sound files clustering into a system of voice authentication of person’s identity has allowed to reduce the probability of a type 2 error (that is identification of a foe as a friend) by 10−3 during the password phrase containing 3 words. The authors constructed an automatic machine for allocation and classification of sound fragments of conjoint speech.
Conclusions. The work offers the numerical algorithm for identification of certain speaker’s speech allowing to synchronize speech segements. The use of the statistical method has allowed to specify the value of the revealed parameters. The conducted research has allowed to construct the automatic machine for allocation and classification of sound fragments on various segements of sound signals. This procedure has been integrated into the structure of the available system of voice authentication and has considerably improved the system’s quality at emergence of the probability of a type 2 error.

Key words

digital processing of signals, numerical methods, biometrics, speech prediction, voice authentication, synchronization of sound fragments of speech

Download PDF
References

1. Agashin O. S., Korelin O. N. Trudy nizhegorodskogo gosudarstvennogo tekhnicheskogo universiteta im. R. E. Alekseeva [Proceedings of Nizhny Novgorod State Technical University named after R.E. Alekseev]. 2012, no. 4 (97), pp. 32–44.
2. Khurgin Ya. I., Yakovlev V. P. Finitnye funktsii v fizike i tekhnike [Finite functions in physics and engineering]. Moscow: Nauka, 1971, 408 p.
3. Stenger F. Springer Series in Computational Mathematics. Springer Verlag, 1993, 565 p.
4. Dodis Y. A., Reyzin L. EUROCRYPT, 2004, April 13, pp. 523–540.
5. Monrose F., Reiter M., Li Q., Wetzel S. Proc. IEEE Symp.on Security and Privacy, 2001, pp. 1–12.
6. Boykov I. V., Ivanov A. I., Kalashnikov D. M. Izvestie vysshikh uchebnykh zavedeniy. Povolzhskiy region. Tekhnicheskie nauki [University proceedings. Volga region. Engineering sciences]. 2015, no. 4 (36), pp. 64–78.
7. Yazov Yu. K., Volchikhin V. I., Ivanov A. I., Funtikov V. A., Nazarov I. G. Neyrosetevaya zashchita personal'nykh biometricheskikh dannykh [Neural-network protection of personal biometric data]. Moscow: Radiotekhnika, 2012, 157 p.
8. Akhmetov B. S., Ivanov A. I., Funtikov V. A., Bezyaev A. V., Malygina E. A. Tekhnologiya ispol'zovaniya bol'shikh neyronnykh setey dlya preobrazovaniya nechetkikh biometricheskikh dannykh v kod klyucha dostupa: monogr. [A technology of large neural networks application for fuzzy biometric data conversion into an access key code: monograph]. Almaty, Kazakhstan: LEM, 2014, 144 p. Available at: http://portal.kazntu.kz/
files/publicate/2014-06-27-11940.pdf

 

Дата создания: 01.07.2016 09:09
Дата обновления: 01.07.2016 10:12