A study on the data augment method considering room transfer functions for acoustic scene classification

Minhan Kim, Seokjin Lee

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Acoustic scene classification is the problem of recognition of sound around our living area. Since people recognize the situation through sound when they can't see, it is very natural that acoustical approach is being made in research that awareness of environment. In this field, research using deep learning method such as CNN is widely used recently. However, this method has the disadvantage that the lower the number of data, the lower the performance. So, in this paper, data augmentation considering acoustical approach-the transfer function of the room-was performed to obtain enough number of data for each classes. To verify this method, we used dataset from DCASE 2018 challenge, which is acoustic scene classification competition. Our augmentation method improved overall f1-score by 0.1 from the state-of-art performance.

Original languageEnglish
Title of host publicationProceedings of the 23rd International Congress on Acoustics
Subtitle of host publicationIntegrating 4th EAA Euroregio 2019
EditorsMartin Ochmann, Vorlander Michael, Janina Fels
PublisherInternational Commission for Acoustics (ICA)
Pages2821-2826
Number of pages6
ISBN (Electronic)9783939296157
DOIs
StatePublished - 2019
Event23rd International Congress on Acoustics: Integrating 4th EAA Euroregio, ICA 2019 - Aachen, Germany
Duration: 9 Sep 201923 Sep 2019

Publication series

NameProceedings of the International Congress on Acoustics
Volume2019-September
ISSN (Print)2226-7808
ISSN (Electronic)2415-1599

Conference

Conference23rd International Congress on Acoustics: Integrating 4th EAA Euroregio, ICA 2019
Country/TerritoryGermany
CityAachen
Period9/09/1923/09/19

Keywords

  • Acoustic Scene Classification
  • Data Augmentation
  • Machine Learning

Fingerprint

Dive into the research topics of 'A study on the data augment method considering room transfer functions for acoustic scene classification'. Together they form a unique fingerprint.

Cite this