University of Twente Student Theses


Towards understanding social interactions through audio signals

Vossebeld, F.R. (2022) Towards understanding social interactions through audio signals.

[img] PDF
Abstract:Everyone has social interactions in their lives. The way these social interactions occur might depend on the kind of person with whom is interacted. Different social relationships can be defined for this. These social relationships can be very different. Think of how the interaction with a stranger is different than with a friend. This research focuses on conversations between different relationships. Audio signal analysis is performed to determine whether the social relationship can be predicted purely on audio features. After this, different audio feature groups are compared to see whether one is more important for the prediction than the other. Then, the importance of the separate acoustic features is looked into. In order to do this, audio features are extracted from labelled conversations for which the social relationship is known. Random Forest performs the best on the data, achieving a balanced accuracy of 0.54 ± 0.06.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Business & IT BSc (56066)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page