S-VERIFY: advanced speaker verification (E!8719) is a Eureka project that will research and develop innovative speaker verification technologies. Lack of robust session variability compensation in state-of-the art system is a major obstacle hindering a wider deployment of speaker verification techniques in secure remote authentication. The innovative approach to session variability compensation developed within the S-VERIFY project, which is thoroughly described in the project plan, will be a radically new solution, superior to existing comparable state-of-the-art solutions for authentication technologies. More information about the project can be found at the official Eureka website.

Authentication based on biometric features of a user is a basic security mechanism used when accessing a particular space, building, or country and for accessing protected information. After 11 September 2001, it has become a major research topic of research among many R&D groups at European, US, and Asian universities and development centers. Speaker verification is the process of automatic validation of the user’s claim that she or he is of certain identity, based on speaker-specific information included in speech waves. It enables access control of various services by voice biometrics. Despite its considerable promise, speaker verification is still not being used on a large-scale basis in public service contexts. Wider acceptance of speaker verification systems will depend on three factors: resilience to circumvention attempts, improved robustness, and more attractive pricing.

IVG10tf100n is a database designed for research in the area of speaker identification and verification, namely for speaker recognition on the basis of digits spoken over the phone. Recording has been repeated once a month with approximately 100 speakers. It was carried out over the telephone network, using a Dialogic CTI card. Samples were recorded on a hard disc in mono PCM format, 16 bits/sample, 8000 samples/second. Each time the caller's name, the calling phone number, two fixed and ten more random sequences of four digits were also recorded. Some of the callers participated in the recording process every month (their voice can be used for system training), and some of them called only once (their voice can be used for faulty identification probability testing).


The dissemination of the results of S-VERIFY project includes the following publications at international scientific conferences and in scientific journals:

Ivan Jokić, Stevan Jokić, Vlado Delić, Zoran Perić (2015) “One Approach to Speaker Modeling for Android Applications”, XII International Conference Society for Electronics, Telecom., Automation and Informatics, ETAI-2015, Ohrid, Macedonia, 24-26.09.2015, ISBN 978- 9989-630-76-7, pp. 8-2.1-8-2.4.

Ivan Jokić, Stevan Jokić, Vlado Delić, Zoran Perić (2015) “Mel-Frequency Cepstral Coefficients as Features for Automatic Speaker Recognition”, 23rd Telecom. Forum, TELFOR-2015, Belgrade, Serbia, 24 26.11.2015, ISBN: 978-1-5090-0054-8, Org: Telecommunications Society, pp. 419-424, http://www.telfor.rs/

Ivan Jokić, Stevan Jokić, Vlado Delić, Zoran Perić (2015) “Influence of a Content of Speech to Accuracy of Automatic Speaker Recognition”, 17th International Conference on Speech and Computer SPECOM-2015, 20-24 September 2015, Athens, Greece, Proceedings Vol. II, University of Patras, ISBN: 978-960-530-160-6, pp. 87-93

Ivan Jokić, Vlado Delić, Stevan Jokić, Zoran Perić (2015) “Automatic Speaker Recognition Dependency on Both the Shape of Auditory Critical Bands and Speaker Discriminative MFCCs”, Advances in Electrical and Computer Engineering, ISSN: 1582-7445 (Accepted 20.11.2015. Paper ID: 3545).

Vlado Delić, Ivan Jokić, Tijana Delić, Slobodan Jošić, Kristina Vakarov, Maja Milišić, Robert Šafer (2015) “Subjective and objective measurement of voice similarity: Recording and analysis of authentic and imitated speaker database“, ETAI2015, Ohrid, Macedonia, 24-26.09.2015, ISBN: 978-9989-630-76-7, pp. 1-2.1-1- 2.2.

Justin, T., Struc, V., Dobrisek, S., Vesnicer, B., Ipšić, I., Mihelič, F. (2015): "Speaker de-identification using diphone recognition and speech synthesis", In: IEEE International Conference on Automatic Face and Gesture Recognition and Workshops.

Vesnicer, B., Žganec-Gros, J., Dobrišek, S., Štruc, V. (2014): "Incorporating Duration Information into I-vector-based Speaker-Recognition systems", In: Odyssey: the Speaker and Language Recognition Workshop, Joensuu, Finland.

Ivan Jokić, Stevan Jokić, Vlado Delić, Zoran Perić (2014) “Towards a Small Intra-Speaker Variability Models”, Elektronika ir Elektrotechnika, ISSN: 1392-1215 (IF2013=0.445), Kaunas University of Technology, Vol. 20, No. 6, pp. 100-103, DOI: 10.5755/j01.eee.20.6.7276, http://www.eejournal.ktu.lt/index.php/elt

Štruc, V., Žganec-Gros, J., Vesnicer, B., Pavešić, N. (2014): "Beyond parametric score normalisation in biometric verification systems", IET Biometrics, vol. 3, no. 2, pp. 62-74.

Edvin Pakoci, Nikša Jakovljević, Branislav Popović, Dragiša Mišković, Darko Pekar (2014) “Speaker Detection Using Phoneme Specific Hidden Markov Models”, SPECOM-2014, Novi Sad, Serbia, 05-09.10.2014, Eds. A. Ronzhin, R. Potapova, V. Delić, Springer LNAI 8773, ISSN: 0302-9743, ISBN: 978-3-319-11580-1, pp. 410-417, DOI: 10.1007/978-3-319-11581-8_51.

Ivan Jokić, Vlado Delić, Stevan Jokić, Zoran Perić (2013) “Influence of the discarding of non-speaker specific model parameters and features to accuracy of automatic speaker recognition”, 2nd International Acoustics and Audio Engineering Conferenc, TAKTONS-2013, 13-16.11.2013, Novi Sad, Serbia, ISBN: 978-86-7892-555-9, pp. 96-99.

Stevan Ostrogonac, Milan Sečujski, Dragan Knežević, Siniša Suzić (2013) “Extraction of Glottal Features for Speaker Recognition”, 9th International Conference on Computational Cybernetics, ICCC-2013, ISBN: 978-1-4799-0060-2, Tihany, Hungary, 8-10.07.2013, Org: Óbuda University, Hungarian Fuzzy Association, pp. 369-373, http://conf.uni-obuda.hu/iccc2013/.