Unlocking the soundscape of coral reefs with artificial intelligence: pretrained networks and unsupervised learning win out DOI Creative Commons
Ben Williams, Santiago Martínez Balvanera, Sarab S. Sethi

et al.

PLoS Computational Biology, Journal Year: 2025, Volume and Issue: 21(4), P. e1013029 - e1013029

Published: April 28, 2025

Passive acoustic monitoring can offer insights into the state of coral reef ecosystems at low-costs and over extended temporal periods. Comparison whole soundscape properties rapidly deliver broad from data, in contrast to detailed but time-consuming analysis individual bioacoustic events. However, a lack effective automated for data has impeded progress this field. Here, we show that machine learning (ML) be used unlock greater soundscapes. We showcase on diverse set tasks using three biogeographically independent datasets, each containing fish community (high or low), cover low) depth zone (shallow mesophotic) classes. supervised train models identify ecological classes sites report unsupervised clustering achieves whilst providing more understanding site groupings within data. also compare different approaches extracting feature embeddings recordings input ML algorithms: indices commonly by ecologists, pretrained convolutional neural network (P-CNN) trained 5.2 million hrs YouTube audio, CNN’s which were task (T-CNN). Although T-CNN performs marginally better across tasks, reveal P-CNN offers powerful tool generating marine as it requires orders magnitude less computational resources achieving near comparable performance T-CNN, with significant improvements indices. Our findings have implications ecology any habitat.

Language: Английский

Temporal patterns in Malaysian rainforest soundscapes demonstrated using acoustic indices and deep embeddings trained on time-of-day estimation DOI Open Access
Yen Yi Loo,

Mei Yi Lee,

Samien Shaheed

et al.

The Journal of the Acoustical Society of America, Journal Year: 2025, Volume and Issue: 157(1), P. 1 - 16

Published: Jan. 1, 2025

Rapid urban development impacts the integrity of tropical ecosystems on broad spatiotemporal scales. However, sustained long-term monitoring poses significant challenges, particularly in regions. In this context, ecoacoustics emerges as a promising approach to address gap. Yet, harnessing insights from extensive acoustic datasets presents its own set such time and expertise needed label species information recordings. Here, study an investigating soundscapes: use deep neural network trained time-of-day estimation. This research endeavors (1) provide qualitative analysis temporal variation (daily monthly) soundscape using conventional ecoacoustic indices embeddings, (2) compare predictive power both methods for estimation, (3) performance supervised classification unsupervised clustering specific recording site, habitat type, season. The study's findings reveal that proposed embeddings exhibit overall comparable performance. article concludes by discussing potential avenues further refinement method, which will contribute understanding across space.

Language: Английский

Citations

0
Bird Diversity of the Dry Chaco: Impacts of Land Use Change on Communities and Soundscapes DOI Open Access
Romina Cardozo, Ricardo B. Machado

Austral Ecology, Journal Year: 2025, Volume and Issue: 50(2)

Published: Feb. 1, 2025

ABSTRACT Agricultural expansion has had negative impacts on biodiversity worldwide. Regions with high human pressure, such as the Dry Chaco in South America, require rapid studies to understand environmental and potential loss. Ecoacoustics been proposed an efficient method for promoting assessment of threatened regions. Using a unique field‐based bird community dataset, we evaluated performance two commonly used acoustic indices (acoustic diversity index complexity index) representing avian richness continuous forest corridors Paraguayan Chaco. Our results from manual identification recordings showed higher species sites (40–61 species) than (22–36 species). In contrast, found no difference between or corridors. Contrary our initial expectation, there was not significant association when considered across all sites. However, partial weak correlation values We argue that habitat fragmentation edge effects might have altered soundscape corridors, favouring activity rather richness, which affects response. study suggests must be cautiously because other variables, besides are involved characterisation (e.g., vocal activity).

Language: Английский

Citations

0
Increased avian bioacoustic diversity without lost profit after planting perennial vegetation in marginal cropland DOI Creative Commons
Adam E. Mitchell,

April Stainsby,

Christy A. Morrissey

et al.

Agriculture Ecosystems & Environment, Journal Year: 2025, Volume and Issue: 388, P. 109663 - 109663

Published: April 5, 2025

Language: Английский

Citations

0
Unlocking the soundscape of coral reefs with artificial intelligence: pretrained networks and unsupervised learning win out DOI Creative Commons
Ben Williams, Santiago Martínez Balvanera, Sarab S. Sethi

et al.

PLoS Computational Biology, Journal Year: 2025, Volume and Issue: 21(4), P. e1013029 - e1013029

Published: April 28, 2025

Passive acoustic monitoring can offer insights into the state of coral reef ecosystems at low-costs and over extended temporal periods. Comparison whole soundscape properties rapidly deliver broad from data, in contrast to detailed but time-consuming analysis individual bioacoustic events. However, a lack effective automated for data has impeded progress this field. Here, we show that machine learning (ML) be used unlock greater soundscapes. We showcase on diverse set tasks using three biogeographically independent datasets, each containing fish community (high or low), cover low) depth zone (shallow mesophotic) classes. supervised train models identify ecological classes sites report unsupervised clustering achieves whilst providing more understanding site groupings within data. also compare different approaches extracting feature embeddings recordings input ML algorithms: indices commonly by ecologists, pretrained convolutional neural network (P-CNN) trained 5.2 million hrs YouTube audio, CNN’s which were task (T-CNN). Although T-CNN performs marginally better across tasks, reveal P-CNN offers powerful tool generating marine as it requires orders magnitude less computational resources achieving near comparable performance T-CNN, with significant improvements indices. Our findings have implications ecology any habitat.

Language: Английский

Citations

0