Stevan Jokić

Facta Universitatis-Series Electronics And Energetics

Mel-Frequency Cepstral Coefficients and Spectrum Based Additional Features in Automatic Speaker Recognition

Ivan Jokić, Stevan Jokić, Vlado Delić, Zoran Perić

https://casopisi.junis.ni.ac.rs/index.php/FUElectEnerg/article/view/13295

The efficiency of the proposed automatic speaker recognizer is evaluated using two speech databases. The feature vector consists of 21 mel-frequency cepstral coefficients (MFCCs), along with up to three additional features derived from the amplitude spectrum. The additional features are calculated based on the logarithm of the energy around the appropriate local maximum in the spectrum, the frequency of that maximum, and the logarithm of the energy of the maximum component in the spectrum across all frames of the observed signal. The speaker identification procedure for a closed set of speakers is tested on the Solo section of the CHAINS database and a speech database with expressed emotions, developed within the S-ADAPT project. The achieved maximum mean recognition accuracies are 97.11% on the CHAINS database and up to 98.72% on the S-ADAPT database.

accuracy, audio recording, human voice, speaker recognition, spectral analysis

M23

AlfaTech

Ethical and Innovative Smartphone-Based Blood Vessel Assessment: Privacy and Data Protection in the "ECG for Everybody" App for Smart Cities

Stevan Jokić, Ivan Jokić, Branislav Gerazov, Nenad Gligorić, Ana Kovačević

https://doi.ub.kg.ac.rs/doi/alfatech25-176j/

The growing emphasis on personalized healthcare within the paradigm of smart cities highlights the need for innovative, accessible solutions that enable early detection and continuous monitoring of cardiovascular health. This study proposes a methodology for assessing blood vessel elasticity (vascular "biological age") through analysis of PPG signals acquired via the mobile application "ECG for Everybody". The approach introduces a dominant PPG beat obtained by averaging signals across the recording and uses deep neural networks and signal processing techniques to estimate vascular health parameters. Initial results validate the efficacy of the approach and support accessible, real-time healthcare solutions for smart city environments.

Photoplethysmography (PPG) analysis, vascular elasticity, biological age, neural networks, healthcare technology, smart cities

M33

AlfaTech

Use of Covariance Matrix in Automatic Speaker Recognition

Ivan Jokić, Stevan Jokic

https://doi.ub.kg.ac.rs/doi/alfatech25-204%D1%98/

One procedure for automatic speaker recognition based on 21 mel-frequency cepstral coefficients as speaker features and a covariance matrix as the speaker model is tested in this paper. Tests are conducted on the Solo part of the CHAINS speech database. Recognition results are compared for two cases: with and without sigmoid function applied to covariance matrix elements. Across the test stages, applying the sigmoid function significantly improves recognition accuracy, increasing the mean recognition accuracy from 87.84% to 94.64%.

Automatic speaker recognition; Mel-Frequency Cepstral Coefficients; Covariance matrix

—

AlfaTech

PPG signal analysis and wavelet selection for feature extraction

Zlatko Radovanović, Stevan Jokić, Ivan Jokić, Branislav Gerazov, Ana Kovačević, Nenad Gligorić

https://doi.ub.kg.ac.rs/doi/alfatech25-220r/

This paper analyzes photoplethysmography (PPG) biosignals and examines wavelet transformation for feature extraction. Biosignals carry important information about physiological mechanisms and health states, and PPG is particularly useful in modern diagnostics. The paper applies artificial intelligence methods and places special focus on selecting wavelets suitable for machine learning tasks based on PPG signal characteristics.

characteristic parameters, diastolic peak, photoplethysmography (PPG), pulse width, machine learning, systolic peak, wavelet transform

M33

AlfaTech

Using machine learning techniques for age prediction based on PPG signal analysis

Mirjana Tomic, Stevan Jokic, Ivan Jokić, Nenad Gligorić, Ana Kovačević, Branislav Gerazov

https://doi.ub.kg.ac.rs/doi/alfatech25-220r/

This paper explores machine learning and neural networks for age prediction based on PPG signals as a non-invasive and cost-effective method for health assessment, especially in cardiovascular medicine. Multiple neural network architectures, activation functions (tanh and ReLU), and preprocessing techniques were tested. Evaluation used MAE and MSE metrics. Results indicate that models with more hidden layers improved performance, reducing errors by about 30% compared to single-layer models, while highlighting the importance of data balance and signal-specific characteristics for further optimization.

PPG signals, machine learning, neural networks, age prediction, data processing, cardiovascular health

M33

Electronics

Mapping Computer Vision Syndrome: An Engineering Problem in Human–Computer Interaction

Viduka Dejan, Vanja Dimitrijević, Dragan Rastovac, Milan Gligorijević, Ana Bašić, Srđan Maričić, and Stevan Jokić

https://www.mdpi.com/2079-9292/13/22/4460

Computer Vision Syndrome (CVS) is highly prevalent but remains relatively understudied. This bibliometric study aims to raise awareness and encourage further research. Data were retrieved from PubMed, Lens, Scopus, and Google Scholar for the period 1 January–31 December 2023, and analyzed using Zotero, VOSviewer, and Microsoft Excel. Out of 893 reviewed papers, 578 were included. The study presents analyses of top authors and publishers, publication trends, sources, and keywords, highlighting increasing research trends and leading countries while emphasizing the need for further research and user awareness.

bibliometric analysis; computer vision syndrome; human–computer interaction; occupational diseases; computer users; computer terminals

M22

Telfor

Decentralized Identities for Enhanced Security in Vehicle-to-Everything

Ana Kovačević, Nenad Gligorić, Stevan Jokić

https://2024.telfor.rs/en/program/

The rise of Vehicle-to-Everything (V2X) communication in Intelligent Transportation Systems enhances road safety and traffic management but introduces security vulnerabilities. This paper proposes a decentralized identity framework using Decentralized Identifiers (DIDs), Verifiable Credentials (VCs), Distributed Ledger Technology (DLT), and Zero-Knowledge Proofs (ZKPs) for secure identity management in V2X systems. The framework improves authentication, reduces centralization risks, and enables privacy-preserving mechanisms with enhanced scalability and resilience.

Vehicle-to-Everything; Decentralized Identifiers; Verifiable Credentials; Zero-Knowledge Proofs; Autonomous Vehicles

M33

IcETRAN

Web ontology design for data and services in CLIMOS project

Nenad Gligorić, Milan Dordevic, Daniel San Martín, Stevan Jokić, Ivan Jokić

https://ieeexplore.ieee.org/document/10192131

This paper presents web ontology design and realization in Protégé for collected data and services in the CLIMOS project. CLIMOS addresses climate change-induced emergence and spread of zoonotic pathogens and relies on large datasets and forecasting services. Web ontologies are used to describe contents and services in machine-readable form and to support annotation, discovery, publication, advertising, and automated service composition, improving data and service reusability and interoperability.

Web ontology, connected data

M33