Razpoznavanje emocij iz glasbenih signalov z uporabo klasičnih metod, globokega učenja in modela wav2vec 2.0 : magistrsko delo

Bulajić, Maša

| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

First page > Show document

Show document

Title:	Razpoznavanje emocij iz glasbenih signalov z uporabo klasičnih metod, globokega učenja in modela wav2vec 2.0 : magistrsko delo
Authors:	ID Bulajić, Maša (Author) ID Donaj, Gregor (Mentor) More about this mentor...
Files:	MAG_Bulajic_Masa_2026.pdf (1,54 MB) MD5: 538A4B9588CDD4C4BE1D8F0467988ABF MAG_Bulajic_Masa_2026.zip (8,78 KB) MD5: 786C9433A527AEB7016390E0DFC32713
Language:	Slovenian
Work type:	Master's thesis/paper
Typology:	2.09 - Master's Thesis
Organization:	FERI - Faculty of Electrical Engineering and Computer Science
Abstract:	Glasba ima pomemben vpliv na človekove emocije. V magistrskem delu raziskujemo metode za avtomatsko razpoznavanje emocij iz glasbenih signalov. Primerjani so trije pristopi: klasične metode strojnega učenja, ki temeljijo na ročno izluščenih akustičnih značilkah, modeli globokega učenja z uporabo konvolucijskih nevronskih mrež (CNN) ter samonadzorovani predtrenirani model wav2vec 2.0. Eksperimenti so izvedeni na podatkovni zbirki DEAM, kjer so zvezne vrednosti valence in arousal pretvorjene v štiri emocionalne razrede. Uspešnost modelov je ocenjena z metrikami točnost, preciznost, priklic in macro F1. Rezultati omogočajo primerjavo pristopov ter pokažejo potencial nadaljnjega izboljšanja.
Keywords:	strojno učenje, globoko učenje, samonadzorovano učenje, razpoznavanje emocij
Place of publishing:	Maribor
Place of performance:	Maribor
Publisher:	[M. Bulajić]
Year of publishing:	2026
Number of pages:	1 spletni vir (1 datoteka PDF (IX, 72 str.))
PID:	20.500.12556/DKUM-97962
UDC:	004.85:78(043.2)
COBISS.SI-ID:	282222083
Publication date in DKUM:	29.05.2026
Views:	158
Downloads:	11
Metadata:
Categories:	KTFMB - FERI
:	Copy citation

Average score:	(0 votes)
Your score:	Voting is allowed only for logged in users.
Share:

Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Licences

License:	CC BY-NC-ND 4.0, Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International

Link:	http://creativecommons.org/licenses/by-nc-nd/4.0/
Description:	The most restrictive Creative Commons license. This only allows people to download and share the work for no commercial gain and for no other purposes.
Licensing start date:	05.05.2026

Secondary language

Language:	English
Title:	Emotion recognition from musical signals using classical methods, deep learning, and the wav2vec 2.0 model
Abstract:	Music has a strong influence on human emotions. This thesis investigates methods for automatic emotion recognition from musical audio signals. Three different approaches are compared: classical machine learning methods based on handcrafted acoustic features, deep learning using convolutional neural networks (CNN), and a self-supervised pretrained model wav2vec 2.0. Experiments are conducted on the DEAM dataset, where continuous valence–arousal annotations are transformed into four emotion classes. The models are evaluated using accuracy, precision, recall and macro F1-score. The results allow comparison of approaches and show the potential for further improvement.
Keywords:	machine learning, deep learning, self-supervised learning, emotion recognition

Comments

Leave comment

You must log in to leave a comment.

Comments (0)

0 - 0 / 0

There are no comments!

Back