diplomsko delo
Tadej Lahovnik (Author), Vili Podgorelec (Mentor)

Abstract

V diplomskem delu smo se poglobili v izdelavo različnih tipov spektrogramov in klasifikacijo slik z uporabo konvolucijskih nevronskih mrež. Zanimalo nas je, ali je možno zanesljivo napovedati žanr zvočnega posnetka glede na spektrogram, ki mu pripada. Tekom razvoja smo ustvarili tri različne tipe spektrogramov. Za vsak tip smo ustvarili ločen klasifikacijski model, nato pa smo iz vseh treh modelov sestavili klasifikacijski ansambel. Tako smo dobili najbolj zanesljive rezultate. Klasifikacijo smo nato ovrednotili s številnimi metrikami, kjer nas je najbolj zanimala sama točnost klasifikacije. Iz matrike zmede smo izčrpali najpogostejše napake pri klasifikaciji.

Keywords

klasifikacija;spektrogram;strojno učenje;nevronske mreže;glasbeni žanri;diplomske naloge;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher: [T. Lahovnik]
UDC: 004.85:004.932(043.2)
COBISS: 130638083 Link will open in a new window
Views: 57
Downloads: 12
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Music genre classification based on the spectrogram of the sound recording
Secondary abstract: In our thesis, we delved into the generation of diverse types of spectrograms and image classification using convolutional neural networks. We were interested in whether it is possible to reliably predict the genre of an audio recording based on its spectrogram. During our development, we created three distinct types of spectrograms. We created a separate classifier model for each type and then built a classifier ensemble from all three models. In this way, we obtained the most reliable results. We then evaluated the classification with several metrics, where we were most interested in the accuracy of the classification. We extracted the most common classification errors from the confusion matrix.
Secondary keywords: classification;spectrogram;machine learning;neural networks;music genre;
Type (COBISS): Bachelor thesis/paper
Thesis comment: Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Informatika in tehnologije komuniciranja
Pages: 1 spletni vir (1 datoteka PDF (X, 36 f.))
ID: 16193677