ArcticSoundsNET: BirdNET embeddings facilitate improved bioacoustic classification of Arctic species

In recent years, deep learning has become a popular solution for processing large ecological monitoring datasets. This rise in use has resulted in global classification models for a variety of data types and taxa, such as BirdNET, which classifies vocalizations of more than 6000 avian species from a...

Full description

Saved in:
Bibliographic Details
Main Authors: Morgan A. Ziegenhorn, Richard B. Lanctot, Stephen C. Brown, Miles Brengle, Shiloh Schulte, Sarah T. Saalfeld, Christopher J. Latty, Paul A. Smith, Nicolas Lecomte
Format: Article
Language:English
Published: Elsevier 2025-12-01
Series:Ecological Informatics
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S1574954125002791
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849233617766580224
author Morgan A. Ziegenhorn
Richard B. Lanctot
Stephen C. Brown
Miles Brengle
Shiloh Schulte
Sarah T. Saalfeld
Christopher J. Latty
Paul A. Smith
Nicolas Lecomte
author_facet Morgan A. Ziegenhorn
Richard B. Lanctot
Stephen C. Brown
Miles Brengle
Shiloh Schulte
Sarah T. Saalfeld
Christopher J. Latty
Paul A. Smith
Nicolas Lecomte
author_sort Morgan A. Ziegenhorn
collection DOAJ
description In recent years, deep learning has become a popular solution for processing large ecological monitoring datasets. This rise in use has resulted in global classification models for a variety of data types and taxa, such as BirdNET, which classifies vocalizations of more than 6000 avian species from acoustic data. These global models can be useful pre-trained models for transfer learning, allowing researchers to more easily develop classifiers specialized to their datasets. However, the development of such models hinges on the availability of comprehensive, high-quality training data, which can be difficult to acquire, produce, and use. We present a novel pipeline for creating training data from a large and unlabeled dataset with minimal human oversight. We used this pipeline and BirdNET as our base model to develop a transfer-learning-based model, ArcticSoundsNET, using acoustic monitoring data from 205 sites across Alaska's Arctic Coastal Plain. We compared performance of ArcticSoundsNET with that of BirdNET to evaluate the effectiveness of our pipeline and success of the new model. We found that the ability of ArcticSoundsNET to detect and classify avian vocalizations in our data greatly exceeded that of BirdNET (AUC ROC = 0.888 for ArcticSoundsNET, AUC ROC = 0.593 for BirdNET). Importantly, our method for developing a training dataset is widely applicable for ecologists who do not have large amounts of labeled data, facilitating the creation of task-specific classification models. Developing such models is an essential step in using large acoustic datasets to support ecological conservation of critical species and habitats.
format Article
id doaj-art-dc4df117c39d491cb2a24eb48e7132cc
institution Kabale University
issn 1574-9541
language English
publishDate 2025-12-01
publisher Elsevier
record_format Article
series Ecological Informatics
spelling doaj-art-dc4df117c39d491cb2a24eb48e7132cc2025-08-20T05:05:28ZengElsevierEcological Informatics1574-95412025-12-019010327010.1016/j.ecoinf.2025.103270ArcticSoundsNET: BirdNET embeddings facilitate improved bioacoustic classification of Arctic speciesMorgan A. Ziegenhorn0Richard B. Lanctot1Stephen C. Brown2Miles Brengle3Shiloh Schulte4Sarah T. Saalfeld5Christopher J. Latty6Paul A. Smith7Nicolas Lecomte8Department of Biology and Centre d'Études Nordiques, University of Moncton, Moncton, New Brunswick E1A 3E9, Canada; Corresponding author.U.S. Fish and Wildlife Service, Migratory Bird Management, Anchorage, AK 99503, USAManomet, Inc., Manomet, MA 02360, USAIndependent ResearcherManomet, Inc., Manomet, MA 02360, USAU.S. Fish and Wildlife Service, Migratory Bird Management, Anchorage, AK 99503, USAU.S. Fish and Wildlife Service, Northern Alaska Fish and Wildlife Field Office, Fairbanks, AK 99701-6237, USANational Wildlife Research Centre, Carleton University, Ottawa, Ontario K1A 0H3, CanadaDepartment of Biology and Centre d'Études Nordiques, University of Moncton, Moncton, New Brunswick E1A 3E9, CanadaIn recent years, deep learning has become a popular solution for processing large ecological monitoring datasets. This rise in use has resulted in global classification models for a variety of data types and taxa, such as BirdNET, which classifies vocalizations of more than 6000 avian species from acoustic data. These global models can be useful pre-trained models for transfer learning, allowing researchers to more easily develop classifiers specialized to their datasets. However, the development of such models hinges on the availability of comprehensive, high-quality training data, which can be difficult to acquire, produce, and use. We present a novel pipeline for creating training data from a large and unlabeled dataset with minimal human oversight. We used this pipeline and BirdNET as our base model to develop a transfer-learning-based model, ArcticSoundsNET, using acoustic monitoring data from 205 sites across Alaska's Arctic Coastal Plain. We compared performance of ArcticSoundsNET with that of BirdNET to evaluate the effectiveness of our pipeline and success of the new model. We found that the ability of ArcticSoundsNET to detect and classify avian vocalizations in our data greatly exceeded that of BirdNET (AUC ROC = 0.888 for ArcticSoundsNET, AUC ROC = 0.593 for BirdNET). Importantly, our method for developing a training dataset is widely applicable for ecologists who do not have large amounts of labeled data, facilitating the creation of task-specific classification models. Developing such models is an essential step in using large acoustic datasets to support ecological conservation of critical species and habitats.http://www.sciencedirect.com/science/article/pii/S1574954125002791Deep learningTransfer learningArcticBirdsBirdNETPassive acoustic monitoring
spellingShingle Morgan A. Ziegenhorn
Richard B. Lanctot
Stephen C. Brown
Miles Brengle
Shiloh Schulte
Sarah T. Saalfeld
Christopher J. Latty
Paul A. Smith
Nicolas Lecomte
ArcticSoundsNET: BirdNET embeddings facilitate improved bioacoustic classification of Arctic species
Ecological Informatics
Deep learning
Transfer learning
Arctic
Birds
BirdNET
Passive acoustic monitoring
title ArcticSoundsNET: BirdNET embeddings facilitate improved bioacoustic classification of Arctic species
title_full ArcticSoundsNET: BirdNET embeddings facilitate improved bioacoustic classification of Arctic species
title_fullStr ArcticSoundsNET: BirdNET embeddings facilitate improved bioacoustic classification of Arctic species
title_full_unstemmed ArcticSoundsNET: BirdNET embeddings facilitate improved bioacoustic classification of Arctic species
title_short ArcticSoundsNET: BirdNET embeddings facilitate improved bioacoustic classification of Arctic species
title_sort arcticsoundsnet birdnet embeddings facilitate improved bioacoustic classification of arctic species
topic Deep learning
Transfer learning
Arctic
Birds
BirdNET
Passive acoustic monitoring
url http://www.sciencedirect.com/science/article/pii/S1574954125002791
work_keys_str_mv AT morganaziegenhorn arcticsoundsnetbirdnetembeddingsfacilitateimprovedbioacousticclassificationofarcticspecies
AT richardblanctot arcticsoundsnetbirdnetembeddingsfacilitateimprovedbioacousticclassificationofarcticspecies
AT stephencbrown arcticsoundsnetbirdnetembeddingsfacilitateimprovedbioacousticclassificationofarcticspecies
AT milesbrengle arcticsoundsnetbirdnetembeddingsfacilitateimprovedbioacousticclassificationofarcticspecies
AT shilohschulte arcticsoundsnetbirdnetembeddingsfacilitateimprovedbioacousticclassificationofarcticspecies
AT sarahtsaalfeld arcticsoundsnetbirdnetembeddingsfacilitateimprovedbioacousticclassificationofarcticspecies
AT christopherjlatty arcticsoundsnetbirdnetembeddingsfacilitateimprovedbioacousticclassificationofarcticspecies
AT paulasmith arcticsoundsnetbirdnetembeddingsfacilitateimprovedbioacousticclassificationofarcticspecies
AT nicolaslecomte arcticsoundsnetbirdnetembeddingsfacilitateimprovedbioacousticclassificationofarcticspecies