Transkripcija kitarske glasbe s konvolucijskimi nevronskimi mrežami

diplomsko delo

Igor Nikolaj Sok (Author), Matija Marolt (Mentor)

Abstract

Avtomatska transkripcija glasbe (AMT) je zelo aktualen problem. Na tem področju se je v zadnjih letih pojavilo ogromno napredkov, a šele pred kratkim so se pojavili poskusi transkripcije kitarske glasbe direktno iz posnetka v tablature, ki pa za razliko od not, za določeno melodijo niso enolične. Cilj diplomske naloge je izdelati program, ki bo sposoben iz poljubnega posnetka kitarske glasbe izluščiti tablature, ki so čim bolj podobne tablaturam, katerim je sledil igralec kitare, ter izvedljive za igranje na kitari. Za ta proces uporabljamo konvolucijske nevronske mreže. Z izgradnjo modela, ki kombinira konvolucijsko ter rekurentno arhitekturo, lahko na področju problema dosežemo zelo dobre rezultate.

Keywords

konvolucijske nevronske mreže;kitara;tablature;transkripcija;računalništvo in informatika;univerzitetni študij;diplomske naloge;

Data

Language:	Slovenian
Year of publishing:	2023
Typology:	2.11 - Undergraduate Thesis
Organization:	UL FRI - Faculty of Computer and Information Science
Publisher:	[I. N. Sok]
UDC:	004.8:004.021:78(043.2)
COBISS:	148004867
Views:	45
Downloads:	11
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Guitar music transcription with convolutional neural networks
Secondary abstract:	Automatic music transcription is a very actual problem in computer vision and multimedia. There have been many advances in the field, but only recently direct transcription from guitar music to tablatures, which, unlike notes, are not uniform, has been attempted. The goal of this diploma thesis is to make a program, that could estimate a playable and accurate tablature from a given recording of guitar music. For that process, we could use convolutional neural networks. By combining a convolutional and recurrent neural network architecture, we can achieve very good results for the task of automatic guitar transcription.
Secondary keywords:	neural networks;convolutional neural networks;guitar;guitar music;tablatures;transcription;computer science;computer and information science;diploma;Nevronske mreže (računalništvo);Kitarska glasba;Računalništvo;Univerzitetna in visokošolska dela;
Type (COBISS):	Bachelor thesis/paper
Study programme:	1000468
Embargo end date (OpenAIRE):	1970-01-01
Thesis comment:	Univ. v Ljubljani, Fak. za računalništvo in informatiko
Pages:	49 str.
ID:	18275614