Modeli strojnog učenja za klasifikaciju glazbenih žanrova na skupu podataka AudioSet

undergraduate thesis

• Title (english): Machine Learning Models for Music Genre Classification on AudioSet Dataset
• Author: Maja Polanec
• Mentor: Marina Bagić Babac (mentor)
• Committee member: Marina Bagić Babac (predsjednik povjerenstva)
• Committee member: Melita Fertalj (član povjerenstva)
• Committee member: Kristijan Poje (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-06-29, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Ova teza se bavi problemom klasifikacije glazbenih žanrova pomoću nadziranog strojnog učenja. Klasificirani žanrovi su pop, rock, hip-hop, vokal, reggae, R&B i techno. Za obuku i testiranje modela koristi se skup podataka AudioSet. Audio značajke iz vremenske i frekvencijske domene su inženjerski oblikovane i proslijeđene 6 različitim modelima: logistička regresija, stablo odlučivanja, slučajne šume, podržani vektori, XGBoost i naivni Bayes. Performanse modela se evaluiraju koristeći različite metrike evaluacije: točnost, matrica konfuzije, F1-ocjena i ROC-krivulja.

Abstract (english)

This thesis deals with the problem of classifying music genres using supervised machine learning. Genres classified are pop, rock, hip-hop, vocal, reggae, R&B and techno. AudioSet dataset is used for training and testing models. Audio features from both time and frequency domain are engineered and fed to 6 different models: logistic regression, decision tree, random forests, support vector machine, XGBoost and naïve Bayes. Model performances are evaluated using different evaluation metrics: accuracy, confusion matrix, F1-score and ROC-curve.

• Keywords: glazbeni žanr; zvučne značajke; strojno učenje: AudioSet; logistička regresija; stablo odluke; slučajna šuma; podržani vektori; XGBoost; naivan Bayes; vremenska domena; frekvencijska domena
• Keywords (english): music genre; audio features; machine learning; AudioSet; logistic regression; decision tree; random forest; support vector machine; XGBoost; naïve Bayes; audio features; time domain; frequency domain
• Language: english
• URN:NBN: urn:nbn:hr:168:355069
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-07-13 10:28:41