Improvement of image analysis/synthesis technologies of acoustic (speech) information for the control, safety and communication systems

Autor: V. Dvoryankin V. Dvoryankin, Nikita S. Dvoryankin, Roman A. Ustinov
Jazyk: English<br />Russian
Rok vydání: 2019
Předmět:
Zdroj: Безопасность информационных технологий, Vol 26, Iss 1, Pp 64-76 (2019)
Druh dokumentu: article
ISSN: 2074-7128
2074-7136
DOI: 10.26583/bit.2019.1.07
Popis: Voice communication has been and remains one of the main ways of human communication and human-machine exchange. Today a construction of new perspective systems of processing and protection of speech information is impossible without modeling of effective mechanisms of speech transformation, creation of speech-like signals with the set properties.For this purpose a unique approach is proposed which deals with transformation of a halftone image of a speech signal spectrogram into a binary one, its subsequent modification in order to solve the problems of protection and processing of speech information, and the possibility of a reverse transition to a halftone image and subsequent synthesis of a new speech-like signal with the desired properties.An improvement of the model of speech formation, making use of the properties of auditory perception and taking into account the features of the formation of binary spectrograms can significantly reduce the amount of speech information without losing its semantic content and recognition and provide an opportunity to use a rich and well-tested arsenal of ways to recognize and process binary and halftone images and a number of other important advantages.The prospects of using image analysis-synthesis technologies in relation to narrow-band sonograms and other kind of images while solving the problems of acoustic steganography, digital noise cleaning and reconstruction of distorted phonograms, audio labeling of significant information, speech compression and restoration are also evaluated.
Databáze: Directory of Open Access Journals