Comparing audio classification using a convolutional- and a recurrent neural network

Keywords

Loading...
Thumbnail Image

Issue Date

2023-01-27

Language

en

Document type

Journal Title

Journal ISSN

Volume Title

Publisher

Title

ISSN

Volume

Issue

Startpage

Endpage

DOI

Abstract

This thesis project will try to evaluate the differences between a CNN and an RNN when used for an audio classification task. A CNN is known to perform well at classifying images, which audio fragments can easily be converted to. Alternatively, an RNN is known to be able to capture (long) time dependencies, which are inherently present in sound as sound is a signal over time. These different advantages may or may not yield positive effects when classifying audio. Differences between the results of the two strategies are interesting to be evaluated and could pertain to the essence of the strategies themselves. The findings can be used to draw conclusions about whether one strategy might be preferable over the other in an audio classification task related to urban sounds.

Description

Citation

Faculty

Faculteit der Sociale Wetenschappen