Prepoznavanje skladatelja klasične glazbe na temelju zapisa skladbi u MIDI formatu

undergraduate thesis

• Title (english): Prediction of classical music composers based on audio pieces stored in MIDI format
• Author: Lara Grgurić
• Mentor: Mirjana Domazet-Lošo (mentor)
• Committee member: Mirjana Domazet-Lošo (predsjednik povjerenstva)
• Committee member: Mario Brčić (član povjerenstva)
• Committee member: Mihael Kovač (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-06-30, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U današnje vrijeme vrlo su popularni softveri za prepoznavanje glazbenog djela i
njenog autora. Tehnologije za takav tip problema koriste baze podataka koje sadrže
goleme količine podataka. Cilj je ovog rada testirati je li moguće predvidjeti autora
neke skladbe kojoj algoritam nema pristup te je li takvo predviđanje moguće ako koristimo
MIDI format glazbenog zapisa. Odabrani pristup rada za ostvarenje tog cilja
je strojno učenje. Koristeći tri različite metode strojnog učenja te tri različite metode
odabira značajki klasificirala sam skladbe prema skladateljima. Dobiveni rezultati navode
na zaključak da je predviđanje autora skladbe uistinu moguće na osnovi MIDI
formata skladbe, što potiče muzikološka istraživanja o uzorcima u djelima najpoznatijih
skladatelja te razvoj novih tehnologija kojima nisu potrebni veliki izvori kako bi
mogli prepoznati autora glazbenog djela.

Abstract (english)

In today’s time, software for music recognition and identifying the author of a
musical piece is very popular. Technologies for such problems utilize databases that
contain vast amounts of data. The goal of this study is to test whether it is possible
to predict the author of a composition to which the algorithm does not have access,
and whether such prediction is possible when using the MIDI format of a music recording.
The chosen approach for achieving this goal is machine learning. By using
three different machine learning methods and three different feature selection methods,
I classified compositions according to their composers. The obtained results lead to the
conclusion that predicting the author of a composition is indeed possible based on the
MIDI format of the composition. This encourages musicological research on patterns
in the works of the most famous composers and the development of new technologies
that do not require extensive resources to identify the author of a musical piece.

• Keywords: MIDI; Klasifikacija glazbe; SVM; XGBoost; Random forest; ANOVA; Informacijska dobit; Odabir značajki na temelju korelacije
• Keywords (english): MIDI; Music classification; SVM; XGBoost; Random forest; ANOVA; Information Gain; Correlation Feature Selection
• Language: croatian
• URN:NBN: urn:nbn:hr:168:979780
• 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 10:05:18