Staff

Research Work

• Speech signal modeling: Autoregressive / Linear Predictive Coding model, fractal-based modeling, non-parametric modeling.
• Speech recognition: deep learning-based recognition, HMM based speech recognition.
• Evaluation of speech quality: estimation of phonation level, acoustic analysis of vocal fold functionality.

Scientific publications

Key publications in the last 5 years:

• K. Kakol, G. Korvel, G. Tamulevičius, B. Kostek. Detecting  Lombard Speech Using Deep Learning Approach, Sensors, 2023, Vol. 1, 315 [https://doi.org/10.3390/s23010315].
• G. A. Melnik-Roy, J. Bernatavičienė, G. Korvel, G. Navickas, G. Tamulevičius, P. Treigys. An Overview of Lithuanian Intonation: A Linguistic and Modelling Perspective, Informatica, 2022, Vol. 33(4), pp. 795-832 [https://doi.org/10.15388/22-INFOR502].

• G. Tamulevičius, G. Korvel, A. B. Yayak, P. Treigys, J. Bernatavičienė, B. Kostek. A Study of Cross-Linguistic Speech Emotion Recognition Based on 2D Feature Spaces, Electronics, 2020, Vol. 9, No. 10, 1725 [https://doi.org/10.3390/electronics9101725].
• P. Treigys, G. Korvel. G. Tamulevičius, J. Bernatavičienė, B. Kostek. Investigating Feature Spaces for Isolated Word Recognition, in Data Science: New Issues, Challenges and Applications, 2020, Vol. 869, pp. 165-181 [https://doi.org/10.1007/978-3-030-39250-5_9].
• G. Tamulevičius, R. Karbauskaitė, G. Dzemyda. Speech emotion classification using fractal dimension-based features, Nonlinear analysis: Modelling and control, 2019, Vol. 24, No. 5, pp. 679-695 [https://doi.org/10.15388/NA.2019.5.1].
• J. Venskus, P. Treigys, J. Bernatavičienė, G. Tamulevičius, V. Medvedev. Real-Time Maritime Traffic Anomaly Detection Based on Sensors and History Data Embedding, Sensors, 2019, Vol.19, No. 17, 3782 [https://doi.org/10.3390/s19173782].

Publications with VU Institute of Data Science and Digital Technologies & Institute of Mathematics and Informatics affiliation (starting 2011).

Presentation at conferences

• G. Tamulevičius, G. Korvel, J. Bernatavičienė, P. Treigys. The Impact of Data Augmentation Based on White Noise for Noise-Robust CNN-Based Speech Recognition. Data Analysis Methods for Software Systems, November 28-30, 2019, Druskininkai [more].
• G. Tamulevičius, A. Valotka. Machine Learning for the Lithuanian language: State-of-the-art and challenges. Machine Learning Applications in the Social Sciences and the Humanities, October 18-19, 2019, Wurzburg, Germany.
• G. Tamulevičius. Will we communicate with robots in Lithuanian? 9th Conference of Lithuanian Terminology "Speech and Terminology Technologies", June 7, 2019, Vilnius.

Scientific and other projects

Horizon 2020 project "National Competence Centres in the framework of EuroHPC (EUROCC)"

• Subject: To establish a single National Competence Centre (NCC) in the area of high-performance computing (HPC) in countries-members. These NCCs will coordinate activities in all HPC-related fields at the national level and serve as a contact point for customers from industry, science, (future) HPC experts, and the general public alike.
• Period: September 2020 - August 2022.

Expert contract in project “Development of a system for the description of studies”

• Subject: Preparation of the descriptor for Informatics Study field.
• Funding: European Social Fund, Lithuania republic state budget.
• Period: September 2019 - September 2020.

COST action CA15225 "Fractional-order systems - analysis, synthesis and their importance for future design"

• Subject: Fractional-order systems have lately been attracting significant attention and gaining more acceptance as generalization to classical integer-order systems. Mathematical basics of fractional-order calculus were laid nearly 300 years ago and since that it has gained deeply rooted mathematical concepts. Today, it is known that many real dynamic systems cannot be described by a system of simple differential equation or of integer-order system. In practice we can encounter such systems in electronics, signal processing, thermodynamics, biology, medicine, control theory, etc. The Action will favour scientific advancement in above mentioned areas by coordinating activities of academic research groups towards an efficient deployment of fractal theory to industry applications. The cooperation of researchers from different institutions will guarantee wide visibility of Action results.
• Period: October 2016  - October 2020.

EU Structural funds project "Development of Lithuanian speech-based services (LIEPA 2)"

• Subject:  Active use in electronic media will become one of the most important survival factors for small languages in the near future. Lithuanian spoken language is classified as so-called non-commercial language and has little access to automatic use of other languages processing tools. Therefore, there is emerging risk that its practical use will become more and more obsolete. Digital products and services created by LIEPA-2 will increase usage of Lithuanian in digital space.
• Period: December 2017 - December 2020.

• Funding: Central project management agency. 
• Contract No.: 02.3.1-CPVA-V-527-01-0001.

EU Structural funds project "Lithuanian speech-based services (LIEPA)"

• Subject: The aim is to create solutions (Lithuanian spoken language (voice) synthesizer and its applications; command and phrase recognizer and its applications; Lithuanian continuous speech recognizer; speech corpus and their applications) by applying modern information and communication technologies and to make it freely accessible for third party employment. 
• Period: January 2013 - July 2015.

• Funding: Information society committee under the Ministry of the economy.
• Contract No.: VP2-3.1-IVPK-12-K-01-001.

Research group R&D project "Development and validation of control by Lithuanian speech unit model for the disabled (GALIA)"

• Subject: The main aim of this project is to design, to develop and to validate the model of voice control unit for disabled
  persons. This device will enable the disabled persons to control necessary functional devices and domestic appliances
  by voice.
• Period: July 2012 - July 2014.
• Funding: Research council of Lithuania.
• Contract No.: MIP-092/2012.

Research service project "Development of Lithuanian isolated word and phrase recognition engine model"

• Subject: The aim is to develop isolated Lithuanian word and phrase recognition model, to prepare recommendations for the development of Lithuanian speech recognition engine.
• Period: December 2012 - March 2013.
• Co-financing partner: Agency for science, innovation and technology.
• Conract No.: APS-580000-3203, JSC „Inogama“.

Research group project "Implementation of Lithuanian isolated word recognition in Field Programmable Gate Array"

• Subject: The main aim of the project is to determine and evaluate the benefits of dynamic time warping applied in isolated word recognition framework implementation by multiple Field Programmable Gate Arrays. Results of the project will widespread dynamic time warping applicability and research. Finally, after the project we will have developed new specialized hardware means for quicker and more precise Lithuanian language isolated word recognition.
• Period: January 2009 - December 2009.
• Funding: State foundation of science and studies.
• Contract No.: T-106/09. Project registration No.: T-09152.

Editorial board and Reviewing

• Reviewer, Informatica.
• Reviewer, IEEE Access.
• Reviewer, Neurocomputing.
• Reviewer, Baltic Journal of Modern Computing.
• Reviewer, Nonlinear Analysis: Modelling and Control.
• Reviewer, IEEE Journal of Biomedical and Health Informatics.
• Reviewer, International Journal of Applied Mathematics and Computer Sciences.

PhD Supervision

Daniel Zakševski

• Subject: Speech signal quality.
• Period of study: 1^st November 2023–31^st October 2027
• Science field: Informatics Engineering (T 007)
• The title of Thesis: Deep learning models for speech enhancement
• Defended:

Monika Danilovaitė

• Subject: Speech signal quality, evaluation of fonation process and vocal fold status.
• Period of study: 1^st October 2020 - 30^th September 2025.
• Science field: Computing (N 009).
• The title of Thesis: Voice quality assessment methodology for the detection of vocal folds disorders.
• Defended: -

Tatjana Liogienė

• Subject: speech emotions.
• Period of study: 1^st October 2012 - 30^th September 2016.
• Science field: Computing (09P).
• The title of the Thesis: Hierarchical classification of speech emotions
• Defended: 28th September 2017.

Teaching Activities

Bachelor degree studies:

• Audio signal processing, Vilnius University, since 2020.
• Front-end development, Vilnius University, since 2018.
• Digital signal processing, Vilnius Gediminas Technical University, since 2012.
• Script programming, Vilnius Gediminas Technical University, 2005 - 2011.

Master degree studies:

• Visualization technologies, Vilnius Gediminas Technical University, 2015.
• Speech signal processing, Vilnius Gediminas Technical University, since 2008.

Science promotion

• Lecture "How sound effects are created. Digital signal processing" in Student for one day programme, 27^th October 2023, Vilnius.
• Presentation of Faculty studies and research in Švenčionys Z. Žemaitis gymnasium.The topic of the lesson: "Digital Signal Processing. The Intersection of Informatics, Mathematics and Engineering“. 18^th October 2023, Švenčionys.
• Presentation of the Mathematics and Informatics faculty study programmes and research activities. Presentation topic: "Will we communicate with robots (in Lithuanian)?". 21^st November 2019, Varėna.
• Presentation of the Institute research activities in workshop "Joint school of cardiology: heart defects". Presentation topic (with J. Kaukėnas): "Computer-based analysis of heart rate regulation". 21^st April 2018, Kaunas region. 

Other activities

• The chair of the Information systems engineering Study program committee (since 2019).
• Member of the Institute of Mathematics and Informatics board (2015-2017).
• Member of IEEE Computer society section (since 2014).
• Member of IEEE Signal processing society section (since 2012).