GAUSSIAN MODELS OF THE HUMAN VOICE IN THE PROBLEMS OF VERIFICATION AND IDENTIFICATION OF THE PERSON FROM SPEECH SIGNALS
Keywords:
verification, identification, mathematical model of voice, speech vectors, training and test kitsAbstract
At this stage of development of data protection technologies, additional protection is being implemented through the analysis of human biometric data. One of the urgent problems in the field of information protection is the introduction of voice recognition systems. These systems make it possible to identify a person's personality by a set of unique characteristics of the voice.
In this paper, an algorithm based on a Gaussian model of the human voice was created and tested. The model is implemented in the Python programming language. This algorithm can later be used as a tool for verification and identification of a person by the speaker's speech signals.
Test conditional vectors from different sets were compared with the reference model. The desired result was confirmed: the more significant the differences between the test vector and the training (reference) model, the lower the similarity coefficient takes. This result suggests that the computational algorithm has been successfully tested and can be used for further tests with real speech vectors.
References
Г.Д. Гефан, В.Б. Иванов, Авторегрессионные векторные модели и их применение в задачах идентификации личности по речевым сигналам // Информационные технологии и проблемы математического моделирования сложных систем. Иркутск: ИИТМ ИрГУПС, 2010. – Вып. 8, с. 74-81
Г.Д. Гефан, В.Б. Иванов, Математическое моделирование систем верификации и идентификации личности по речевым сигналам // Информационные системы контроля и управления в промышленности и на транспорте. Иркутск: ИрГУПС, 2010. – Вып. 17, с. 24-31
Picone J. Signal modeling techniques in speech recognition – 1993 [Электронный ресурс]. – Режим доступа: http://aprodeus.narod.ru
Bimbot F., Magrin-Chagnolleau I., Mathan L. Second Order Statistical Measures For Text Independent Speaker Identification // Speech Communica-tion. – 1995. – 17(1-2). – P. 177-192.
Burges C.J.C. A tutorial on Support Vector Machines for pattern recog-nition // Data Mining and Knowledge Discovery. – 1998. – V. 2. – No 2. – P. 1-47.
Montacie C., Le Floch J.-L. AR-vector models for free-text speaker recognition // Proceeding of ICSLP. – 1992. – V.1. – P. 611-614.
Rabiner L.R. A tutorial on hidden Markov models and selected applica-tions in speech recognition // Proc. IEEE. – 1989. – 77(2). – P. 257-286.
