Effective speech-based student authentication procedure in distance learning

Nowadays it is almost impossible to find a university that does not provide its students with online courses or correspondence education. Due to various advantages, distance learning has attracted more and more people in recent years. As a result, some of the requirements that this educational format has to satisfy have been included in legislation systems. The necessity to authenticate students remotely is presented as a compulsory procedure in many official documents. When teachers are deprived of face-to-face contact with their students, there is a need to find an appropriate way to verify their personality distantly. In this paper we propose the speech-based student authentication procedure which operates with some acoustic characteristics extracted from voice signals. However, there is one crucial question related to the classification model providing high performance. It is almost impossible for the online systems to vary classifiers and determine the most effective one while interacting with a user. Therefore, to increase the reliability of our proposal we elaborated some classification schemes based on collective decision making to take into account predictions of different classifiers. To prove the effectiveness of this approach, we used a number of multi-lingual corpora (German, English, Japanese). According to the results obtained, a high level of speaker recognition was achieved (up to 100 % of F-score values). The developed algorithmic schemes provide a guaranteed level of effectiveness and might be used as a reliable alternative to the occasional choice of a classification model.