Institutional Repository of Vilnius University
Kalbos emociju˛ atpažinimas naudojant mašinini˛ mokymąsi /
Abstract
dc:descriptionThe aim of this bachelor's thesis is to create a program designed to facilitate the analysis of voice emotions and provide the user with an interactive environment for viewing audio segments and identifying emotions. The program automatically divides the audio recording into segments of fixed duration, creates their mel spectrograms, and applies a deep learning model for emotion classification. The model was trained using the RAVDESS dataset, and its performance was evaluated using several classification quality metrics. Experimental results showed that the model reliably recognizes certain emotion classes, especially clearly expressed emotions such as calm and happy, while distinguishing between similar emotions remains more difficult. The program includes the ability to interactively view segment spectrograms, listen to individual audio fragments, and analyze the distribution of emotions over time. The prototype meets functional and non-functional requirements and can be used for learning and experimental research purposes, and in the future, the system could be improved using larger data sets and more advanced models.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Černiauskas, Arnas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD277689461&prefLang=en_US