Augmenting training dataset with real noise environment for classification of exotic and endangered animal sound

Kyungdeuk Ko, Chulwon Choi, Hanseok Ko

Research output: Contribution to journalConference articlepeer-review


The interest in ecosystem conservation is growing with many remedial policies sought and implemented. In particular, the public recently has perceived awareness that it is essential to eradicate exotic invasive species and protect endangered species because they destroy the food chain and the ecosystem. With the recent advancement of deep learning-based technologies, several studies on acoustic signal classification have taken place, but the studies on animal sound classification considering actual noise environment are still scarce. In this paper, we propose a data augmentation scheme for training by setting up an experimental environment that includes the sounds of exotic invasive species, endangered species, and realistic environmental noise. With the proposed augmentation scheme incorporating more real environment under various SNR conditions, we construct the training set, validation set, and test set to classify the animal sounds. For validation, we compare the performances of various deep learning-based methods used in signal classification tasks by employing the constructed dataset.

Original languageEnglish
JournalProceedings of the International Congress on Acoustics
Publication statusPublished - 2022
Event24th International Congress on Acoustics, ICA 2022 - Gyeongju, Korea, Republic of
Duration: 2022 Oct 242022 Oct 28

Bibliographical note

Funding Information:
This work was supported by Korea Environment Industry & Technology Institute(KEITI) through Exotic Invasive Species Management Program, funded by Korea Ministry of Environment(MOE)(2021002280004)

Publisher Copyright:
© ICA 2022.All rights reserved


  • Animal sound classification
  • Bio-acoustic signal classification
  • Deep learning

ASJC Scopus subject areas

  • Mechanical Engineering
  • Acoustics and Ultrasonics


Dive into the research topics of 'Augmenting training dataset with real noise environment for classification of exotic and endangered animal sound'. Together they form a unique fingerprint.

Cite this