Kamil Kamiński

Phonoscopy is a forensic technique aimed at assessing the authenticity of audio recordings, reconstructing and transcribing the content of conversations, as well as identifying and verifying speakers through speech analysis. One of the most complex challenges in forensic phonoscopic analysis is speaker verification. In Poland, the linguo-acoustic method is employed, which involves examining a range of individual speech characteristics (e.g., accentuation patterns, breathing, or vocabulary choices) alongside the assessment of measurable parameters (including fundamental frequency, pitch, and signal energy). The analysis thus encompasses the processes of phonation and articulation, as well as the prosodic aspects involved in the generation of speech signals. The objective of the presented study was to evaluate the usefulness of both analytical and emergent speech features in speaker verification within forensic proceedings and to develop verification methods that offer substantial support for forensic phonoscopy experts. The analyses were conducted using both analytical physical and behavioral features, as well as emergent features generated by appropriately trained deep neural networks – namely, x-vectors. The combination of both feature types with a proprietary similarity scoring procedure yielded highly promising results: for analytical features, only one instance of false verification was recorded out of 9,900 trials, along with two false rejections out of 100 trials. In the case of x-vectors, a flawless performance was achieved, accompanied by a significantly wide separation margin. The developed procedure may offer considerable support in expert reasoning within the domain of forensic phonoscopy.

Considering the possibility of using terahertz technology in non-destructive testing for military applications, the Consortium consisting of: Military Institute of Armament Technology (Leader), Institute of Optoelectronics from the Military University of Technology, Air Force Institute of Technology and Technology Transfer undertook, within the framework of the project “Innovative methods of terahertz diagnostics in selected key applications in military and security” funded by the National Centrum for Research and Development, the development of test stands for such testing. The project included the development of prototypes of two diagnostic test stands using the transmission and reflection methods for testing. In the course of the project, a number of dielectric composite materials used in military applications were tested. A description of the test stands and their operating principle as well as selected test results are presented in this paper.terahertz radiation, non-destructive testing, composites,

Given the growing threat posed by the widespread availability of unmanned aircraft systems (UASs), which can be utilised for various unlawful activities, the need for a standardised method to evaluate the effectiveness of systems capable of detecting, tracking, and identifying (DTI) these devices has become increasingly urgent. This article draws upon research conducted under the European project COURAGEOUS, where 260 existing drone detection systems were analysed, and a methodology was developed for assessing the suitability of C-UASs in relation to specific threat scenarios. The article provides an overview of the most commonly employed technologies in C-UASs, such as radars, visible light cameras, thermal imaging cameras, laser range finders (lidars), and acoustic sensors. It explores the advantages and limitations of each technology, highlighting their reliance on different physical principles, and also briefly touches upon the legal implications associated with their deployment. The article presents the research framework and provides a structural description, alongside the functional and performance requirements, as well as the defined metrics. Furthermore, the methodology for testing the usability and effectiveness of individual C-UAS technologies in addressing specific threat scenarios is elaborated. Lastly, the article offers a concise list of prospective research directions concerning the analysis and evaluation of these technologies.

In this paper, we report our use of a hardware encoder-based synchronization method for a fast terahertz time-domain spectroscopy raster scanner built with the commercially available TeraFlash Smart platform. We describe the principles of our method, including our incorporation of synchronization signals from various devices included in the scanner. We also describe its implementation in a microcontroller with a dedicated counter. By such means, a fast scanning mode was obtained, which was 35 times faster than a traditional step-by-step approach. To validate the proposed synchronization method, we carried out measurements using the USAF 1951 resolution test and a fiberglass plate with a set of intentionally introduced defects. Our results confirmed that the TDS scanner with the developed synchronization method was able to capture high-quality images with resolutions as high as those obtained using traditional step-by-step scanning, but with significantly reduced scanning times.

This paper presents a voice biometrics system implemented in a web application as part of a two-factor authentication (2FA) user login. The web-based application, via a client interface, runs registration, preprocessing, feature extraction and normalization, classification, and speaker verification procedures based on a modified Gaussian mixture model (GMM) algorithm adapted to the application requirements. The article describes in detail the internal modules of this ASR (Automatic Speaker Recognition) system. A comparison of the performance of competing ASR systems using the commercial NIST 2002 SRE voice dataset tested under the same conditions is also presented. In addition, it presents the results of the influence of the application of cepstral mean and variance normalization over a sliding window (WCMVN) and its relevance, especially for voice recordings recorded in varying acoustic tracks. The article also presents the results of the selection of a reference model representing an alternative hypothesis in the decision-making system, which significantly translates into an increase in the effectiveness of speaker verification. The final experiment presented is a test of the performance achieved in a varying acoustic environment during remote voice login to a web portal by the test group, as well as a final adjustment of the decision-making threshold.

This article presents the Automatic Speaker Recognition System (ASR System), which successfully resolves problems such as identification within an open set of speakers and the verification of speakers in difficult recording conditions similar to telephone transmission conditions. The article provides complete information on the architecture of the various internal processing modules of the ASR System. The speaker recognition system proposed in the article, has been compared very closely to other competing systems, achieving improved speaker identification and verification results, on known certified voice dataset. The ASR System owes this to the dual use of genetic algorithms both in the feature selection process and in the optimization of the system’s internal parameters. This was also influenced by the proprietary feature generation and corresponding classification process using Gaussian mixture models. This allowed the development of a system that makes an important contribution to the current state of the art in speaker recognition systems for telephone transmission applications with known speech coding standards.

We built a high-speed TDS setup with the use of electronically controlled optical sampling (ECOPS), which can measure up to 1600 terahertz pulses per second. The movement of the sample was provided by two fast-speed motorized linear stages constituting the gantry. We developed a flat-bar-based metal marker approach for the synchronization of continuous line-by-line scans. We carefully compared the performance of the terahertz reflection time-domain spectroscopy (TDS) scanner operating in a slow point-by-point and a one-hundred-times faster line-by-line imaging scheme. We analyzed images obtained for both schemes for a uniform metallic breadboard with holes, as well as a glass composite sample with defects. Although the measurement time was reduced by 100 times in terms of the line-by-line scheme, the overall performance in both schemes was almost identical in terms of the defects’ sizes, shapes and locations. The results proved that the proposed ECOPS TDS system can provide uniform and extremely fast scanning without any deterioration in image quality.