| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

Show document Help

Title:Samodejna napoved vokalne privlačnosti z uporabo globokega učenja in razložljive umetne inteligence : magistrsko delo
Authors:ID Bulajić, Anja (Author)
ID Kačič, Zdravko (Mentor) More about this mentor... New window
ID Ntalampiras, Stavros (Comentor)
Files:.pdf MAG_Bulajic_Anja_2026.pdf (1,75 MB)
MD5: 95D8106A45881D054C39BFD91ECF237A
 
.zip MAG_Bulajic_Anja_2026.zip (15,53 MB)
MD5: 480F89A94FBA8DF5E10D21F61E74B5C6
 
Language:Slovenian
Work type:Master's thesis/paper
Typology:2.09 - Master's Thesis
Organization:FERI - Faculty of Electrical Engineering and Computer Science
Abstract:V magistrskem delu je obravnavan problem samodejnega napovedovanja vokalne privlačnosti na podlagi govornega signala. Uporabljena je bila podatkovna zbirka CocoNut-Humoresque z zaznavnimi ocenami poslušalcev, izraženimi kot povprečna mnenjska ocena (MOS). Razvita sta bila klasični regresijski model na osnovi ročno izbranih akustičnih značilk in globoki model CNN–BiLSTM na podlagi mel-spektrogramov. Uspešnost modelov je bila ovrednotena z regresijskimi metrikami, pri čemer je bil analiziran vpliv porazdelitve ocen na napake napovedovanja. Z uporabo metode Grad-CAM je bila dodatno izvedena analiza razložljivosti modela in identifikacija ključnih spektro-časovnih vzorcev, povezanih z zaznano vokalno privlačnostjo.
Keywords:vokalna privlačnost, CNN–BiLSTM, mel-spektrogram, Grad-CAM
Place of publishing:Maribor
Place of performance:Maribor
Publisher:A. Bulajić
Year of publishing:2026
Number of pages:1 spletni vir (1 datoteka PDF (IX, 66 str.))
PID:20.500.12556/DKUM-97714 New window
UDC:004.85(043.2)
COBISS.SI-ID:282227459 New window
Publication date in DKUM:29.05.2026
Views:181
Downloads:9
Metadata:XML DC-XML DC-RDF
Categories:KTFMB - FERI
:
Copy citation
  
Average score:(0 votes)
Your score:Voting is allowed only for logged in users.
Share:Bookmark and 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:Automatic prediction of vocal attractiveness using deep learning and explainable artificial intelligence
Abstract:This master’s thesis addresses the problem of automatic prediction of vocal attractiveness from speech signals. The CocoNut-Humoresque dataset with perceptual listener ratings expressed as Mean Opinion Score (MOS) was used. A classical regression model based on hand-crafted acoustic features and a deep CNN–BiLSTM model operating on mel-spectrograms were developed. Model performance was evaluated using standard regression metrics, including an analysis of prediction errors across different MOS ranges. Explainable Artificial Intelligence techniques, specifically Grad-CAM, were applied to interpret model decisions and to identify the most relevant spectro-temporal speech patterns associated with perceived vocal attractiveness.
Keywords:vocal attractiveness, CNN–BiLSTM, mel-spectrogram, Grad-CAM


Comments

Leave comment

You must log in to leave a comment.

Comments (0)
0 - 0 / 0
 
There are no comments!

Back
Logos of partners University of Maribor University of Ljubljana University of Primorska University of Nova Gorica