V.A. Aladinskiy – Ph. D. (Eng.), Professor, 

LTD «STC» (St. Petersburg)

E-mail: awa2810@yandex.ru

S.V. Kuzminskiy – Ph. D. (Eng.), Lecturer, 

Cherepovets highest military engineering college of radio electronics (Cherepovets) V.D. Lebedev – Сadet, 

Cherepovets highest military engineering college of radio electronics (Cherepovets) E-mail: valeralebedal@mail.ru

In the known method, selection is based on the analysis of the values of the normalized autocorrelation function (ACF) r={r0,r1,r2,…,rn,…,rN-1} of the input digital stream (DS) of the channel level (CL), where yn is the n-th character (bit), N is the number of characters in the DS y. There is a method in which on the results of the allocation of extrema in the ACF r with an  interval Δr equal to the block volume Nб of the DS of the CL, the decision is made about the presence in the y block structure and speech message.

The disadvantage of these theoretical results is a decrease in the quality of selection of digital streams with voice messages with an increase in the proportion of digital streams with a block structure containing data. Improving the quality of selection can be achieved in the analysis of AKF r based on the calculation of the vector of linear prediction coefficients (LPC) aM={a1,a2,…,am,…,aM}. The value From the results of the calculation of the vector values aM by the Levinson-Durbin iterative algorithm, it was found that for digital streams containing speech messages, the average EM ср (for 200 tests) is significantly higher than the similar values for digital streams containing texts or images. In addition, the vector of LPC aM of the DS of the CL with speech message has a global minimum. The criterion for making a decision on the selection of the DS of the CL containing a speech message, it is advisable to consider the fulfillment of two conditions: the value of EM > 0,2; in the vector of LPC aM has a global minimum am_min for m={Nб}.

Based on the algorithm of Levinson-Durbin and based on the results of the experiment was developed algorithm of selection of the DS of the CL, containing the voice message.

1. Timofeev D.I., Tavalinskiy D.A., Chubatyy D.N. Analiz parametrov nizkoskorostnykh koderov rechi v usloviyakh strukturnoy i parametricheskoy neopredelennosti. Naukoemkie tekhnologii. 2011. № 8. S. 4–9. [in Russian]
2. Venttsel' E.S., Ovcharov L.A. Teoriya veroyatnostey i ee inzhenernye prilozheniya. M.: Nauka. 1988. ISBN 5-02-013748-0. [in Russian]
3. Patent RF № RU 2610285 S1. Sposob raspoznavaniya protokolov nizkoskorostnogo kodirovaniya / Aladinskiy V.A., Kuz'minskiy S.V., Smirnov P.L., Chubatyy D.N. Byull. № 4. Prioritet ot 08.02.2017. [in Russian]
4. Wai Chu C. Speech coding algorithms: Foundation and evolution of standardized coders. Hoboken: Wiley-Interscience. 2003. 
5. Aladinskiy V.A., Kuz'minskiy S.V. Analiz tsifrovykh potokov na vykhodakh vokoderov, primenyaemykh na zarubezhnykh liniyakh radiosvyazi diapazona vysokikh chastot. Uspekhi sovremennoy radioelektroniki. 2015. № 7. S. 71–74. [in Russian]
6. Matematicheskiy entsiklopedicheskiy slovar'. Gl. red. Yu.V. Prokhorov. M.: Bol'shaya rossiyskaya entsiklopediya. 1995. Reprintnoe  izdanie. M.: Sov. entsiklopediya. 1988. [in Russian]