The developed prototype of an intelligent security system for high-security buildings received the award for the best innovation among the research organizations at the 4th Slovenian Innovations. The system learns the normal behaviour patterns of an employee, recognizes the unusual behaviour, such as thefts, sabotages, staff negligence and insubordination, in real-time. It informs the supervisor with an alarm message in natural language, a graphical explanation and video recordings.
COBISS.SI-ID: 21347879
To uncover criminal organizations, the police make use of various special methods and instruments, like tapping and recording telephone calls. Later, in trial hearing, these pieces of evidence are frequently doubted of. Moreover, it is very difficult to prove that the recorded voice belongs to the defendant and not some other person. A great help in this process comes from the state-of-the-art methods for forensic speaker recognition.
COBISS.SI-ID: 8967761
We tested the ability of humans and machines (data mining techniques) to assign stress to Slovene words. The goal of finding good machine-made models for stress assignment was set by applying new methods and by making use of a known theory about rules for stress assignment in Slovene. The upgraded data mining methods outperformed expert-defined rules on practically all subtasks, thus showing that data mining can more than compete with humans when constructing formal knowledge about stress assignment is concerned.
COBISS.SI-ID: 22516263
The impact of clause and intraclausal coordination detection to dependency parsing of Slovene is examined. New methods based on machine learning and heuristic rules are proposed for clause and intraclausal coordination detection. They were included in a new dependency parsing algorithm, PACID. For evaluation, Slovene dependency treebank was used. At parsing, 6.4 % and 7.1 % relative error reduction was achieved, compared to the dependency parsers MSTP and Malt, respectively.
COBISS.SI-ID: 22911783
Quality of the voice communication channel and the perceived voice quality is very important during voice recording in the forensic investigation. In this paper we present voice quality measurements based on Personal Evaluation of Speech Quality algorithm (PESQ) in fixed and mobile telephony. The results of the PESQ measurements will enable development of the slovenian language speech corpora for further research and evaluation of the forensic speaker recognition systems.
COBISS.SI-ID: 22104871