Music Augmentation Usage in Music Genre Classification Models
Name
Raivo Kasepuu
Abstract
In this master's thesis, research is carried out to find out how the music augmentation affects the accuracy of music genre classification models. To solve the posed problem, the accuracy of the standard model with the original non-augmented dataset is compared with the accuracy of the models created using different augmented datasets. The master thesis uses GTZAN music dataset and music augmentations are observed on MFCC coefficients-based music genre classification models. As a result of the research, it was found that music genre classification models trained on augmented music datasets are more accurate than models trained on non-augmented music datasets.
Graduation Thesis language
Estonian
Graduation Thesis type
Master - Conversion Master in IT
Supervisor(s)
Anna Aljanaki
Defence year
2023