DISCO: A deep learning ensemble for uncertainty-aware segmentation of acoustic signals

Thomas Colligan; Kayla Irish; Douglas J. Emlen; Travis J. Wheeler

doi:10.1371/journal.pone.0288172

DISCO: A deep learning ensemble for uncertainty-aware segmentation of acoustic signals

Thomas Colligan
, Kayla Irish
, Douglas J. Emlen
, Travis J. Wheeler

Division of Biological and Biomedical Sciences

Research output: Contribution to journal › Article › peer-review

1 Scopus citations

Abstract

Recordings of animal sounds enable a wide range of observational inquiries into animal communication, behavior, and diversity. Automated labeling of sound events in such recordings can improve both throughput and reproducibility of analysis. Here, we describe our software package for labeling elements in recordings of animal sounds, and demonstrate its utility on recordings of beetle courtships and whale songs. The software, DISCO, computes sensible confidence estimates and produces labels with high precision and accuracy. In addition to the core labeling software, it provides a simple tool for labeling training data, and a visual system for analysis of resulting labels. DISCO is open-source and easy to install, it works with standard file formats, and it presents a low barrier of entry to use.

Original language	English
Article number	e0288172
Pages (from-to)	e0288172
Journal	PLoS ONE
Volume	18
Issue number	7 July
DOIs	https://doi.org/10.1371/journal.pone.0288172
State	Published - Jul 2023

Funding

Funding:Wealsoacknowledgefundingfromthe NationalInstituteofGeneralMedicalSciences (NIGMS),NationalInstitutesofHealth(NIH) GM132600,andtheDivisionofIntegrative OrganismalSystems(IOS),NationalScience Foundation(NSF)2015907.Thefundershadno roleinstudydesign,datacollectionandanalysis, decisiontopublish,orpreparationofthe manuscript. We also acknowledge funding from the National Institute of General Medical Sciences (NIGMS), National Institutes of Health (NIH) GM132600, and the Division of Integrative Organismal Systems (IOS), National Science Foundation (NSF) 2015907. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors wish to thank Nathan Barton and Camille Thomas-Bulle for their guidance in acquiring, interpreting, and labeling beetle recordings. We are grateful for the use of the GSCC cluster at the University of Montana.

Funder number
2015907
GM132600

Keywords

Animals
Deep Learning
Uncertainty
Reproducibility of Results
Acoustics
Whales
Vocalization, Animal

Access to Document

10.1371/journal.pone.0288172

Cite this

@article{828844be62774685903fa0b6e4a60120,

title = "DISCO: A deep learning ensemble for uncertainty-aware segmentation of acoustic signals",

abstract = "Recordings of animal sounds enable a wide range of observational inquiries into animal communication, behavior, and diversity. Automated labeling of sound events in such recordings can improve both throughput and reproducibility of analysis. Here, we describe our software package for labeling elements in recordings of animal sounds, and demonstrate its utility on recordings of beetle courtships and whale songs. The software, DISCO, computes sensible confidence estimates and produces labels with high precision and accuracy. In addition to the core labeling software, it provides a simple tool for labeling training data, and a visual system for analysis of resulting labels. DISCO is open-source and easy to install, it works with standard file formats, and it presents a low barrier of entry to use.",

keywords = "Animals, Deep Learning, Uncertainty, Reproducibility of Results, Acoustics, Whales, Vocalization, Animal",

author = "Thomas Colligan and Kayla Irish and Emlen, \{Douglas J.\} and Wheeler, \{Travis J.\}",

note = "Copyright: {\textcopyright} 2023 Colligan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.",

year = "2023",

month = jul,

doi = "10.1371/journal.pone.0288172",

language = "English",

volume = "18",

pages = "e0288172",

journal = "PLoS ONE",

issn = "1932-6203",

publisher = "Public Library of Science",

number = "7 July",

}

TY - JOUR

T1 - DISCO

T2 - A deep learning ensemble for uncertainty-aware segmentation of acoustic signals

AU - Colligan, Thomas

AU - Irish, Kayla

AU - Emlen, Douglas J.

AU - Wheeler, Travis J.

N1 - Copyright: © 2023 Colligan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

PY - 2023/7

Y1 - 2023/7

N2 - Recordings of animal sounds enable a wide range of observational inquiries into animal communication, behavior, and diversity. Automated labeling of sound events in such recordings can improve both throughput and reproducibility of analysis. Here, we describe our software package for labeling elements in recordings of animal sounds, and demonstrate its utility on recordings of beetle courtships and whale songs. The software, DISCO, computes sensible confidence estimates and produces labels with high precision and accuracy. In addition to the core labeling software, it provides a simple tool for labeling training data, and a visual system for analysis of resulting labels. DISCO is open-source and easy to install, it works with standard file formats, and it presents a low barrier of entry to use.

AB - Recordings of animal sounds enable a wide range of observational inquiries into animal communication, behavior, and diversity. Automated labeling of sound events in such recordings can improve both throughput and reproducibility of analysis. Here, we describe our software package for labeling elements in recordings of animal sounds, and demonstrate its utility on recordings of beetle courtships and whale songs. The software, DISCO, computes sensible confidence estimates and produces labels with high precision and accuracy. In addition to the core labeling software, it provides a simple tool for labeling training data, and a visual system for analysis of resulting labels. DISCO is open-source and easy to install, it works with standard file formats, and it presents a low barrier of entry to use.

KW - Animals

KW - Deep Learning

KW - Uncertainty

KW - Reproducibility of Results

KW - Acoustics

KW - Whales

KW - Vocalization, Animal

UR - https://www.scopus.com/pages/publications/85165932836

UR - https://www.mendeley.com/catalogue/c6ed103b-76a1-3b50-95a2-7132bc906ec8/

U2 - 10.1371/journal.pone.0288172

DO - 10.1371/journal.pone.0288172

M3 - Article

C2 - 37494341

AN - SCOPUS:85165932836

SN - 1932-6203

VL - 18

SP - e0288172

JO - PLoS ONE

JF - PLoS ONE

IS - 7 July

M1 - e0288172

ER -

DISCO: A deep learning ensemble for uncertainty-aware segmentation of acoustic signals

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this