Find research datasets worth reusing

opencc-zeroFeb 2024View details →

Acoustic features as a tool to visualize and explore marine soundscapes: Applications illustrated using marine mammal Passive Acoustic Monitoring datasets

Passive Acoustic Monitoring (PAM) is emerging as a solution for monitoring species and environmental change over large spatial and temporal scales. However, drawing rigorous conclusions based on acoustic recordings is challenging, as there is no consensus over which approaches, and indices are best suited for characterizing marine and terrestrial acoustic environments. Here, we describe the application of multiple machine-learning techniques to the analysis of a large PAM dataset. We combine pre-trained acoustic classification models (VGGish, NOAA & Google Humpback Whale Detector), dimensionality reduction (UMAP), and balanced random forest algorithms to demonstrate how machine-learned acoustic features capture different aspects of the marine environment. The UMAP dimensions derived from VGGish acoustic features exhibited good performance in separating marine mammal vocalizations according to species and locations. RF models trained on the acoustic features performed well for labelled sounds in the 8 kHz range, however, low and high-frequency sounds could not be classified using this approach. The workflow presented here shows how acoustic feature extraction, visualization, and analysis allow for establishing a link between ecologically relevant information and PAM recordings at multiple scales. The datasets and scripts provided in this repository allow replicating the results presented in the publication.

opencc-zeroJun 2022View details →

Estimating the abundance of the critically endangered Baltic Proper harbour porpoise (Phocoena phocoena) population using passive acoustic monitoring

Knowing the abundance of a population is a crucial component to assess its conservation status and develop effective conservation plans. For most cetaceans, abundance estimation is difficult given their cryptic and mobile nature, especially when the population is small and has a transnational distribution. In the Baltic Sea, the number of harbour porpoises (Phocoena phocoena) has collapsed since the mid-20th century and the Baltic Proper harbour porpoise is listed as Critically Endangered by the IUCN and HELCOM; however, its abundance remains unknown. Here, one of the largest ever passive acoustic monitoring studies was carried out by eight Baltic Sea nations to estimate the abundance of the Baltic Proper harbour porpoise for the first time. By logging porpoise echolocation signals at 298 stations during May 2011-April 2013, calibrating the loggers' spatial detection performance at sea, and measuring the click rate of tagged individuals, we estimated an abundance of 71-1,105 individuals (95% CI, point estimate 491) during May-October within the population's proposed management border. The small abundance estimate strongly supports that the Baltic Proper harbour porpoise is facing an extremely high risk of extinction, and highlights the need for immediate and efficient conservation actions through international cooperation. It also provides a starting point in monitoring the trend of the population abundance to evaluate the effectiveness of management measures and determine its interactions with the larger neighbouring Belt Sea population. Further, we offer evidence that design-based passive acoustic monitoring can generate reliable estimates of the abundance of rare and cryptic animal populations across large spatial scales.

zenodo40/100

Unlabeled AnuraSet: A dataset for leveraging unlabeled data in machine learning models for passive acoustic monitoring

The Unlabeled AnuraSet (U-AnuraSet) is an extension of the original AnuraSet dataset. It consists of soundscape recordings from passive acoustic monitoring conducted in Brazil. The recording sites are identical to those in the original AnuraSet. Each site comprises 2,666 one-minute raw audio files of unlabeled data. The U-AnuraSet is publicly available to encourage machine learning researchers to explore innovative methods for leveraging unlabeled data in the training of models aimed at solving problems such as anuran call identification. If you find the Unlabeled AnuraSet useful for your research, please consider citing it as follows: Cañas, J.S., Toro-Gómez, M.P., Sugai, L.S.M., et al. A dataset for benchmarking Neotropical anuran calls identification in passive acoustic monitoring. Sci Data 10, 771 (2023). https://doi.org/10.1038/s41597-023-02666-2

opencc-by-4.0May 2024View details →

opencc-zeroFeb 2023View details →

Data for: Large-scale long-term passive-acoustic monitoring reveals spatiotemporal activity patterns of boreal bats

The distribution ranges and spatio-temporal patterns in the occurrence and activity of boreal bats are yet largely unknown due to their cryptic lifestyle and lack of suitable and efficient study methods. We approached the issue by establishing a permanent passive-acoustic sampling setup spanning the area of Finland to gain an understanding on how latitude affects bat species composition and activity patterns in northern Europe. The recorded bat calls were semi-automatically identified for three target taxa; Myotis spp., Eptesicus nilssonii or Pipistrellus nathusii and the seasonal activity patterns were modeled for each taxa across the seven sampling years (2015–2021). We found an increase in activity since 2015 for E. nilssonii and Myotis spp. For E. nilssonii and Myotis spp. we found significant latitude -dependent seasonal activity patterns, where seasonal variation in patterns appeared stronger in the north. Over the years, activity of P. nathusii increased during activity peak in June and late season but decreased in mid season. We found the passive-acoustic monitoring network to be an effective and cost-efficient method for gathering bat activity data to analyze spatio-temporal patterns. Long-term data on the composition and dynamics of bat communities facilitates better estimates of abundances and population trend directions for conservation purposes and predicting the effects of climate change.

zenodo40/100

Passive acoustic monitoring applied to black-and-white ruffed lemurs (Varecia variegata) in Ranomafana National Park, Madagascar

Data accompanying the paper: "An integrated passive acoustic monitoring and deep learning pipeline applied to black-and-white ruffed lemurs (\textit{Varecia variegata}) in Ranomafana National Park, Madagascar" Fieldwork was conducted at Mangevo (21.3833S, 47.4667E), an isolated and undisturbed forest location within Ranomafana National Park (RNP), located in southeastern Madagascar, during the period of May to July 2019. To facilitate passive acoustic monitoring, we deployed a total of two SongMeter SM4 devices (manufactured by Wildlife Acoustics) and two Swift units (provided by the Cornell Yang Center for Conservation Bioacoustics). The placement of these recorders was strategically chosen within the central regions of known subgroups, ensuring a minimum distance of 300 meters between each device. The SongMeter devices operated at a sampling rate of 48 kHz, while the Swift units operated at 32 kHz, respectively, enabling comprehensive audio data collection throughout the study period. We provide the audio data (.wav) used to train and test our neural network classifier along with the corresponding labelled text files (.data). Files provided <ul> <li>Test_Audio.zip -- contains (.wav) testing audio files</li> <li>Test_Annotations.zip -- contains (.svl) manually annotated testing files which can be read in using Sonic Visualiser or by parsing the XML file in Python or another programming language. Load in the audio file into Sonic Visualiser and then drag-and-drop the corresponding .svl file.</li> <li>Training_Audio_batch_x.zip -- several .zip files were created to simplify downloading. There are 10 batches, each is roughly 4GB. Each batch contains (.wav) training audio files</li> <li>Training_Annotations.zip -- contains (.svl) manually annotated training files which can be read in using Sonic Visualiser or by parsing the XML file in Python or another programming language. Load in the audio file into Sonic Visualiser and then drag-and-drop the corresponding .svl file.</li> <li>model_weights_tensorflow.hdf5 -- the Tensorflow model. Load the model using: model = tf.keras.models.load_model(model_filepath) note that the model expects a three channel input as explained in the research article.</li> </ul>

opencc-by-nc-sa-4.0May 2023View details →

publicSep 2025View details →

Data from: Optimizing passive acoustic monitoring (PAM) for Biodiversity Studies: using species-area relationship (SAR) to predict species richness

Open the record for dataset details and reuse information.

publicFeb 2024View details →

Acoustic features as a tool to visualize and explore marine soundscapes: Applications illustrated using marine mammal Passive Acoustic Monitoring datasets

Open the record for dataset details and reuse information.

publicFeb 2023View details →

Data for: Large-scale long-term passive-acoustic monitoring reveals spatiotemporal activity patterns of boreal bats

Open the record for dataset details and reuse information.

publicJun 2022View details →

Estimating the abundance of the critically endangered Baltic Proper harbour porpoise (Phocoena phocoena) population using passive acoustic monitoring

Open the record for dataset details and reuse information.

opencc-by-4.0Jun 2016View details →

Comparing distribution of harbour porpoises (Phocoena phocoena) derived from satellite telemetry and passive acoustic monitoring

Data used for publication in Plos One. Two excel files. The satellite_filtered_data is the filtered satellite positions used for MaxEnt modelling in R. The CPOD_data_PPH is the raw C-POD data expressed here as porpoises positive hours (PPH) and can easily be converted to porpoise positive days (PPD).

opencc-by-4.0Sep 2023View details →

A labelled dataset of the loud calls of four vertebrates collected using passive acoustic monitoring in Malaysian Borneo

Passive acoustic monitoring data collection We collected data using first generation Swift autonomous recording units (ARUs) (Koch et al. 2016) with a microphone sensitivity of −44 (+/−3) dB re 1 V/Pa. The microphone frequency response was not measured but is assumed to be flat (+/− 2 dB) in the frequency range 100 Hz to 7.5 kHz. The analog signal was amplified by 40 dB and digitized (16-bit resolution) using an analog-to-digital converter (ADC) with a clipping level of −/+ 0.9 V.  We collected acoustic data from one primary conservation area in Sabah, Malaysia: Danum Valley Conservation Area (with 11 recording units from March to July 2018). Danum Valley covers an area of roughly 440 km², and is characterized by lowland dipterocarp forest. Unlike many tropical forest regions, this area is considered 'aseasonal' due to its lack of clearly differentiated wet and dry seasons (Walsh and Newbery 1999). In Danum Valley, the ARUs recorded at a sampling rate of 16 kHz. All recordings were saved in waveform audio (.wav) format, with files of 2-hr duration. We affixed each recording unit to trees approximately 2-m above the ground and recorded continuously over 24 hours. We set the units on a 750 m grid structure, and preliminary field tests indicate that with these recording settings the detection range of gibbon vocalizations is ~ 400 m.   Acoustic data processing We randomly chose approximately 500 h of recordings from Danum Valley Conservation Area to use to create a training dataset. We used a band-limited energy detector (BLED) to identify potential sounds of interest in the gibbon frequency range. For the BLED detector, we convert the 2-hr recordings into a spectrogram using a 1,600-point (100 ms) Hamming window (3 dB bandwidth = 13 Hz) with 0% overlap and a 2,048-point DFT, with the "seewave" package (Sueur et al. 2008). We then filtered the spectrogram to focus on the desired frequency range, specifically 0.5–1.6 kHz for Northern grey gibbons. For each unique time window in the recording, we determined the total energy across frequency bins which gave a single value for every 100 ms interval. Utilizing the "quantile" function in base R, we established the threshold to delineate signal from noise. Preliminary tests with varied quantile values revealed that the 15th quantile led to optimized recall for our target signal. This approach resulted in 1,439 unique sound events. The sound events were then annotated by a single observer (DJC) using a custom-written function in R to visualize the spectrograms into the following categories: great argus pheasant (Argusianus argus) long and short calls (Clink et al. 2021), helmeted hornbills (Rhinoplax vigil), rhinoceros hornbills (Buceros rhinoceros), female gibbons (Hylobates funereus) and a catch-all “noise” category.  Update Version 5 and later Includes labeled test clips from a second conservation area, Maliau Basin Conservation Area, Sabah, Malaysia recorded during August 2019. The ARUs recorded at a sampling rate of 16 kHz. All recordings were saved in waveform audio (.wav) format, with files of 2-hr duration. We affixed each recording unit to trees approximately 2-m above the ground and recorded continuously over 24 hours.   References Clink, D. J., Groves, T., Ahmad, A. H., & Klinck, H. (2021). Not by the light of the moon: Investigating circadian rhythms and environmental predictors of calling in Bornean great argus. PloS one, 16(2), e0246564. Koch, R., Raymond, M., Wrege, P., & Klinck, H. (2016). SWIFT: A small, low-cost acoustic recorder for terrestrial wildlife monitoring applications. In North American Ornithological Conference (p. 619). Washington, D.C. Sueur, J., Aubin, T., & Simonis, C. (2008). Seewave: a free modular tool for sound analysis and synthesis. Bioacoustics, 18, 213–226. Walsh, R. P., & Newbery, D. M. (1999). The ecoclimatology of Danum, Sabah, in the context of the world’s rainforest regions, with particular reference to dry periods and their impact. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 354(1391), 1869–83. https://doi.org/10.1098/rstb.1999.0528 Webb, C. O., & Ali, S. (2002). Plants and vegetation of the Maliau Basin Conservation Area, Sabah, East Malaysia. Final Report to Maliau Basin Management Committee.

Thyolo alethe (Chamaetylas choloensis) calls for passive acoustic monitoring

Data accompanying the paper: "Passive Acoustic Monitoring and Transfer Learning" Please cite this dataset as: <blockquote> Dufourq, Emmanuel and Batist, Carly and Foquet, Ruben and Durbach, Ian. (2022). Passive Acoustic Monitoring and Transfer Learning. BioRxiv doi:  </blockquote> This dataset contains approximately 10 hours of audio that contained calls of the vulnerable Thyolo Alethe (Chamaetylas choloensis). The audio data was collected in the Mount Mulanje Biosphere Reserve, Malawi using 10 Audiomoths. The sampling rate was set to 32,000Hz and the recordings were obtained over five days in November 2020. A larger dataset exists. The annotations files are in (.svl) format which is compatible with SonicVisualiser (https://www.sonicvisualiser.org/). Each audio file has a corresponding .svl file. Each .svl has segments of audio that were manually annotated as either ''thyolo-alethe" (presence class) or "noise" (absence class) -- this dataset can be used to train a binary classification model. The audio files are provided in "Audio.zip" and the manually verified annotation in "Annotations.zip".

Pin-tailed whydah (Vidua macroura) calls for passive acoustic monitoring

Data accompanying the paper: "Passive Acoustic Monitoring and Transfer Learning" Please cite this dataset as: <blockquote> Dufourq, Emmanuel and Batist, Carly and Foquet, Ruben and Durbach, Ian. (2022). Passive Acoustic Monitoring and Transfer Learning. BioRxiv doi:  </blockquote> This dataset contains approximately 6 hours of audio that contained calls of the pin-tailed whydah (Vidua macroura). The audio data was collected in the Intaka Island Nature Reserve in Cape Town, South Africa using 1 Audiomoth. The sampling rate was set to 48,000Hz and the recordings were obtained over four days in January 2021. A larger dataset exists. The annotations files are in (.svl) format which is compatible with SonicVisualiser (https://www.sonicvisualiser.org/). Each audio file has a corresponding .svl file. Each .svl has segments of audio that were manually annotated as either ''thyolo-alethe" (presence class) or "noise" (absence class) -- this dataset can be used to train a binary classification model. The audio files are provided in "Audio.zip" and the manually verified annotation in "Annotations.zip".

Black-and-white ruffed lemur (Varecia variegata) calls for passive acoustic monitoring

Data accompanying the paper: "Passive Acoustic Monitoring and Transfer Learning" Please cite this dataset as: <blockquote> Dufourq, Emmanuel and Batist, Carly and Foquet, Ruben and Durbach, Ian. (2022). Passive Acoustic Monitoring and Transfer Learning. BioRxiv doi:  </blockquote> This dataset contains approximately 60 hours of audio that contained calls of the critically endangered Black-and-white ruffed lemur (Varecia variegata). The audio data was collected in a sub-humid rainforest site (Mangevo) in the southeast of Ranomafana National Park in Madagascar using 2 Swift recorders (Cornell Center for Conservation Bioacoustics). The sampling rate was set to 48,000Hz and the recordings were collected intermittently between May 2019 and November 2020. A larger dataset exists and further recordings will be released. The annotations files are in (.svl) format which is compatible with SonicVisualiser (https://www.sonicvisualiser.org/). Each audio file has a corresponding .svl file. Each .svl has segments of audio that were manually annotated as either ''thyolo-alethe" (presence class) or "noise" (absence class) -- this dataset can be used to train a binary classification model. The audio files are provided in "Audio.zip" and the manually verified annotation in "Annotations.zip".

Hainan gibbons (Nomascus hainanus) calls for passive acoustic monitoring

This dataset extends an existing one (10.5281/zenodo.3991714). Data accompanying the paper: "Passive Acoustic Monitoring and Transfer Learning" Please cite this dataset as: <blockquote> Dufourq, Emmanuel and Batist, Carly and Foquet, Ruben and Durbach, Ian. (2022). Passive Acoustic Monitoring and Transfer Learning. BioRxiv doi:  </blockquote> This dataset contains approximately 10 hours of audio that contained calls of the critically endangered Hainan gibbons (Nomascus hainanus). The audio data was collected in the Bawangling National Nature Reserve, Malawi using 8 Song Meter SM3 recorders. The sampling rate was set to 9,600Hz and the recordings were collected between March to August 2016. The annotations files are in (.svl) format which is compatible with SonicVisualiser (https://www.sonicvisualiser.org/). Each audio file has a corresponding .svl file. Each .svl has segments of audio that were manually annotated as either ''gibbon" (presence class) or "no-gibbon" (absence class) -- this dataset can be used to train a binary classification model. The audio files are provided in "Audio.zip" and the manually verified annotation in "Annotations.zip".

opencc-zeroJun 2022View details →

dryad36/100

Data from: Passive acoustic monitoring provides reliable under-estimates of population size and longevity in wild Savannah Sparrows

Many breeding birds produce conspicuous sounds, providing tremendous opportunities to study free-living birds through acoustic recordings. Traditional methods for studying population size and demographic features depend on labour-intensive field research. Passive acoustic monitoring provides an alternative method for quantifying population size and demographic parameters, but this approach requires careful validation. To determine the accuracy of passive acoustic monitoring for estimating population size and demographic parameters, we used autonomous recorders to sample an island-living population of Savannah Sparrows (Passerculus sandwichensis) over a six-year period. Using the individually distinctive songs of males, we estimated male population size as the number of unique songs detected in the recordings. We analyzed songs across six years to estimate birth year, death year, and longevity. We then compared the estimates to field data in a blind analysis. Estimates of male population size through passive acoustic monitoring were, on average, 72% of the true male population size, with higher accuracy in lower-density years. Estimates of demographic rates were lower than true values by 29% for birth year, 23% for death year, and 29% for longevity. This is the first investigation to estimate longevity with passive acoustic monitoring, and adds to a growing number of studies that have used passive acoustic monitoring to estimate population size. Although passive acoustic monitoring under-estimated true population parametersfeatures, likely due to the high similarity among many male songs, our findings suggest that autonomous recorders can provide reliable estimates of population size and demographic characteristicslongevity in a wild songbird.

opencc-by-nc-sa-4.0Aug 2021View details →

Pin-tailed whydah and Cape robin-chat calls for passive acoustic monitoring

We provide the audio data (.wav) used to train neural network classifiers along with the corresponding labelled files (.svl). The .svl files are natively read using Sonic Visualiser (https://www.sonicvisualiser.org/) but can directly be read using Python as these are XML files. This is a three class classification dataset. The recordings were obtained using an AudioMoth which was placed at one location in Intaka Island Nature Reserve, Cape Town, South Africa. The recorder was attached to a tree at approximately 1.5 meters from the ground. The sampling rate was set to 48000Hz with a bit rate of 768kbps. The recordings took place in January 2021. While further recordings exist we only provide a small subset here. Additional data can be requested. Files provided Audio.zip - contains audio files (.wav) Annotations.zip - contains the corresponding labels (.svl) for Sonic Visualiser Class description CRC: calls of the Cape robin-chat (Cossypha caffra) PTW: calls of the pin-tailed whydah (Vidua macroura) NOISE: any sound event that does not contain a Cape robin-chat or pin-tailed whydah call   Parts of this data was used in two MSc dissertations: <ul> <li>"Acoustic Data Augmentation for Small Passive Acoustic Monitoring Datasets", Aime Nshimiyimana, African Centre of Excellence in Data Science (ACE-DS) of the University of Rwanda, College of Business and Economics</li> <li>"Pre-training neural networks on Xeno-Canto and eBird for bioacoustic classification models", Mikwa Boris Tamanjong, African Centre of Excellence in Data Science (ACE-DS) of the University of Rwanda, College of Business and Economics</li> </ul>

dryad36/100

Passive acoustic monitoring indicates Barred Owls are established in northern coastal California and management intervention is warranted

Barred Owls (Strix varia) have recently expanded westward from eastern North America, contributing to substantial declines in Northern Spotted Owls (Strix occidentalis caurina). Passive acoustic monitoring (PAM) represents a potentially powerful approach for tracking range expansions like the Barred Owl's, but further methods development is needed to ensure that PAM-informed occupancy models meaningfully reflect population processes. Focusing on the leading edge of the Barred Owl range expansion in coastal California, we used a combination of PAM data, GPS-tagging, and active surveys to (1) estimate breeding home range size, (2) identify patterns of vocal activity that reflect resident occupancy, and (3) estimate resident occupancy rates. Mean breeding season home range size (452 ha) was reasonably consistent with the size of cells (400 ha) sampled with autonomous recording units (ARUs). Nevertheless, false-positive acoustic detections of Barred Owls frequently occurred within cells not containing an activity center such that site occupancy estimates derived using all detected vocalizations (0.61) were unlikely to be representative of resident occupancy. However, the proportion of survey nights with confirmed vocalizations (VN) and the number of ARUs within a sampling cell with confirmed vocalizations (VU) were indicative of Barred Owl residency. Moreover, the false positive error rate could be reduced for occupancy analyses by establishing thresholds of VN and VU to define detections, although doing so increased false negative error rates in some cases. Using different thresholds of VN and VU, we estimated resident occupancy to be 0.29–0.44, which indicates that Barred Owls have become established in the region but also that timely lethal removals could still help prevent the extirpation of Northern Spotted Owls. Our findings provide a scalable framework for monitoring Barred Owl populations throughout their expanded range and, more broadly, a basis for converting site occupancy to resident occupancy in PAM programs.

opencc-zeroJun 2023View details →

dryad36/100

Open‐source workflow approaches to passive acoustic monitoring of bats

<ol> <li>The affordability, storage, and power capacity of compact modern recording hardware has evolved passive acoustic monitoring (PAM) of animals and soundscapes into a non-invasive, cost-effective tool for research and ecological management and is particularly effective for bats and toothed whales that consistently echolocate. The use of PAM at large scales hinges on effective automated detectors and species classifiers which, combined with distance sampling approaches, have enabled species abundance estimation of toothed whales. But standardized, user-friendly, and open-access automated detection and classification workflows are in demand for this key conservation metric to be realized for bats.</li> <li>We used the PAMGuard toolbox including its new deep learning classification module to test the performance of four open-source workflows for automated analyses of acoustic datasets from bats. Each workflow used a different initial detection algorithm followed by the same deep learning classification algorithm and was evaluated against the performance of an expert manual analyst.</li> <li>Workflow performance depended strongly on the signal-to-noise ratio and detection algorithm used: the full deep learning workflow had the best classification accuracy (≤67%) but was computationally too slow for practical large-scale bat PAM. Workflows using PAMGuard's detection module or triggers onboard an SM4BAT or AudioMoth accurately classified up to 47%, 59% and 34%, respectively, of calls to species. Not all workflows included noise sampling critical to estimating changes in detection probability over time, a vital parameter for abundance estimation. The workflow using PAMGuard's detection module was 40 times faster than the full deep learning workflow and missed as few calls (recall for both ~0.6), thus balancing computational speed and performance. </li> <li>We show that complete acoustic detection and classification workflows for bat PAM data can be efficiently automated using open-source software such as PAMGuard and exemplify how detection choices, whether pre- or post-deployment, hardware or software-driven, affect the performance of deep learning classification and the downstream ecological information that can be extracted from acoustic recordings. In particular, understanding, and quantifying detection/classification accuracy and the probability of detection are key to avoid introducing biases that may ultimately affect the quality of data for ecological management. </li> </ol>

opencc-zeroAug 2023View details →