Browsing by Author "Rybalskyi, Oleh"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    ОБЩАЯ МЕТОДИКА ВЫСОКОЭФФЕКТИВНОЙ ДИАГНОСТИКИ СЛЕДОВ МОНТАЖА В ЦИФРОВЫХ ФОНОГРАММАХ. GENERAL METHODOLOGY OF HIGH-EFFICIENCY DIAGNOSTICS OF TRACKS OF EDITING IN DIGITAL PHONOGRAMS
    (2019) Рыбальский, Олег Владимирович; Rybalskyi, Oleh; Соловьев, Виктор Иванович; Solovіov, Viktor; Журавель, Вадим Васильевич; Zhuravel, Vadym
    Into the complete set of tools necessary for the judicial technical examination of materials and apparatus of the audio recording, the programs and methodologies (tool) for diagnostic researches of authenticity of phonograms are obligatorily included. Diagnostics of authenticness of phonograms unites researches of originality (i.e. to priority) of phonogram and presence (or absence) in it the tracks of editing. Presently for the record of the phonograms the digital apparatus of the audio recording is mainly used, and editing is produced with the use of digital processing of the signals fixed in phonograms. Therefore there is an insistent necessity of creation of tool for the exposure of tracks of digital treatment (tracks of the digital editing) in phonograms. It is offered by us, along with the instruments worked out before, to apply the neuron networks of the deep learning for creation of the expert tool intended for the exposure of tracks of digital treatment in phonograms. Preliminary researches are completed, and the first version of the program of such high-efficiency system is built. An aim of the study is the creation of general methodology of examining of the authenticity of digital phonograms, based on the tool worked out by authors. In the construction of such system it is suggested to come from that for editing phonogram it is possible from different pre-product – phonograms taped on the different apparatus of the audio recording, and phonograms, taped on one apparatus. For the exposure of tracks arising up at such types of editing, it is necessary to use a different tool, that allows considerably to shorten labour intensiveness and time of examination. For diagnostics of editing, made from the records taped on different apparatus, it is expedient to use the tool built on the basis of fractal approach, in particular, the automated tool is “Fractal”. For diagnostics of editing, made from the records taped on one apparatus of the audio recording, it is expedient to use the tool built on the basis of neuron network of the deep learning. Both types of tool are worked out by authors. Such approach allows to bring down labour intensiveness and time of examining.
  • Thumbnail Image
    Item
    ПУТИ ПОВЫШЕНИЯ ЭФФЕКТИВНОСТИ КРИМИНАЛИСТИЧЕСКОЙ ИДЕНТИФИКАЦИИ АППАРАТУРЫ ЦИФРОВОЙ ЗВУКОЗАПИСИ. WAYS OF ENHANCING THE EFFECTIVENESS OF FORENSIC IDENTIFICATION OF DIGITAL AUDIO RECORDING EQUIPMENT
    (2019) Рыбальский, Олег Владимирович; Rybalskyi, Oleh; Соловьев, Виктор Иванович; Soloviov, Viktor; Журавель, Вадим Васильевич; Zhuravel, Vadym
    The article deals with the issues of building a toolkit for diagnosing the originality of digital soundtracks and identification of digital audio recording equipment. One of the last variants of creation of the expert toolkit intended for identification of digital recording equipment is the “Fractal” toolkit, but its efficiency is reduced because identification signs are allocated from signals of pauses in the language information (pauses between words) recorded on a phonogram. The purpose of the article is to improve the expert toolkit for identification of sound recording equipment based on the fractal approach to the signals of its own noises. It is suggested to use the separation of hardware noises not only from pauses, but also from a mixture of signals with the entire duration of the phonogram. At the same time, the volume of information from which the identification features of the equipment are allocated increases significantly. Such separation of the own noise signals ensures the increased efficiency of the identification system. The proposed methods of separation and processing of self-similar structures from phonograms, used as identification features in the examination. One of the proposed methods is based on the separation of self-similar structures of the phonogram’s own noises along its entire length from their mixture with language and sound environment signals. The second method provides for mandatory building of error curves of the first and second types when developing the system. When building them, it is necessary to use a large amount of data that can be obtained from a limited number of phonograms by dividing them into separate sections of different lengths in automatic mode. It has been established that for phonograms lasting more than 20 seconds, the error curves are stable, in other words, they practically do not change. The degree of closeness of parameters of self-similar structures, separated from the compared phonograms, is offered to define as the module of distance between the normalized fractal curves, received at measurements of fractal characteristics. The proposed methods allow increasing considerably the efficiency of establishing the originality of soundtracks and identification of digital recording equipment during the expertise.