Preliminary signal processing methods used to create new tools to examine materials and digital sound recording means are described. It is shown that using information redundancy when creating a ...training base for deep learning neural networks used for such examination increases speaker identification efficiency based on voice characteristic parameters by about 15%. It is shown that the proposed processing methods enable speaker identification based on phonograms that are 1 second long.
A method for developing an expert toolkit for forensic identification of digital video recording equipment and digital cameras is considered. The need for developing this toolkit is substantiated. ...The authors proposed to perform the equipment identification using the statistical characteristics of its intrinsic noise, extracted from digital images recorded with the above equipment. The features and primary sources of the above noises in digital images are described. A wavelet analysis based on the Haar wavelet was used for their extraction and processing. The final result of the forensic examination was obtained using deep learning neural networks. The results of using the developed equipment identification system showed its high efficiency.
The authors use the model of a deep learning neuron network to substantiate and verify principal applicability of such network to create a highly effective examination tool for phonogram digital ...processing detection. An experiment was conducted on more than 100,000 experimental fragments of unprocessed and processed phonogram pauses obtained automatically. The obtained dependences showed that when the probability threshold for correct binary classification of pauses is more than 0.55, it is possible to construct a highly effective examination tool.
The authors analyze the models based on deep learning neural networks on the basis of the general approach to pauses and speech signals as different types of audio information fixed in a phonogram, ...different in some characteristics. It is shown that such an approach allows generating the learning database with the use of the general for pauses and signals of speech methods of preliminary processing of information. This provides a higher level of unification of network learning methods intended for solution of various examination problems.
The paper discusses a fundamental approach to creating a system for detecting editing points in digital videograms and proposes the corresponding method. The above approach is the end-product of ...research on identifying digital image recording equipment based on its intrinsic noises recorded on digital media. The authors have found that the method for detecting and localizing editing points in videograms should rely on using functions that describe the dynamics of errors in identifying adjacent frames and the dynamics of the absolute difference in signal levels between the color signals of the two frames in the inspected videogram. The authors have proposed decomposing a signal using the Haar wavelet to obtain these functions. They also have demonstrated that one should implement the system using deep learning neural networks, ensuring high forensic expertise reliability.
Розглянуто процес досліджень, розробки та перевірки придатності програми і методики для ідентифікації диктора за параметрами сигналів мовлення. У розробці програми використано фрак- тальний метод ...представлення ідентифікаційних ознак. Іл.: 2. Бібліогр.: 19 найм.