Listening forward: emerging roles of bioacoustics in ecology, evolution, and conservation
Open accessThe DOI matches its Crossref record; 216 of the 216 references it cites were found in a registry or an index; 39 of the 189 sources its claims cite were checked for support.
Abstract
Bioacoustics is increasingly shifting from a mostly descriptive pursuit to one that can anticipate ecological change. Recent innovations—from autonomous recording units and edge-computing sensors to speech-inspired feature extraction and machine-learning techniques like transfer learning, unsupervised discovery, and explainable AI—are transforming the study of animal communication. These advances let us work at scales previously difficult to imagine. Automated species recognition, individual identification, and even tracking cultural evolution over decades are now within reach. Entire ecosystem soundscapes can be mapped with unprecedented resolution. Looking ahead, global listening networks, adaptive acoustic indices, and live biodiversity dashboards seem increasingly realistic. We may soon build digital models that simulate communication networks under future scenarios. Closer integration with genomics, physiology, and robotics could link vocal traits to their genetic, physiological, and ecological drivers. Challenges remain, including data governance, acoustic privacy, and equitable access to the planet’s sonic heritage. Bioacoustics may be on the way to becoming a predictive, integrative science — one particularly well suited to monitoring, interpreting, and helping safeguard life’s communication systems in a rapidly changing world.
Overview
CoChat Summary
TL;DR
This narrative review traces bioacoustics from early sound-visualization experiments through the digital revolution to today's AI-driven, sensor-network era, arguing the field is shifting from descriptive to predictive science. It surveys current innovations in recording hardware, acoustic indices, and machine learning, then outlines emerging frontiers—global listening networks, biodiversity dashboards, ecosystem "digital twins"—alongside ethical, data-governance, and reproducibility challenges.
Key findings
- Autonomous Recording Units (ARUs), edge computing, and cloud-linked systems (e.g., ARBIMON) are shifting monitoring from episodic surveys to continuous, near-real-time ecological intelligence, with citizen-science projects like Soundscapes to Landscapes deploying hundreds of recorders.
- Acoustic indices are useful but imperfect: meta-analyses show they correlate only moderately with traditional biodiversity metrics (r ≈ 0.33) and behave unpredictably across habitats, taxa, and microphone setups.
- Machine learning now achieves near-human species recognition (CNNs at 89–97% accuracy; >90% with transfer learning), individual identification at 95–99% accuracy across taxa, and semi-supervised pseudo-labeling at >97% sensitivity and >96% specificity; as few as 20–30 labeled examples can suffice for fine-tuning.
- Emerging hardware shows promise: field-programmable neuromorphic chips reached >94% detection rates in noisy real-world conditions, and quantum-enhanced transformers reportedly exceeded state-of-the-art acoustic scene classification by over 5%.
- Data infrastructure lags: fewer than one in four bioacoustics papers deposit recordings in open repositories; one repository holds 52 years and >100 TB of sound, ocean observatories add hundreds of GB daily (cumulative >600 TB), and NOAA houses over 200 TB of passive ocean acoustic data.
Methods
- Design: a narrative review (author's stated intent: "In this review, I trace how bioacoustics reached this turning point and outline where it is headed next"), organized into historical roots, current innovations, applications across scales, and future directions.
- Data/sample: no primary data; the paper synthesizes published literature and cites numerous studies, with two summary tables (Table 1, Table 2) referenced but not reproduced in the text.
- Analysis: qualitative synthesis of cited works, including reported quantitative results from prior studies (e.g., accuracy ranges, correlation coefficients, storage volumes).
Limitations
- Stated in the paper: detection is inherently imperfect because species differ in vocal amplitude, calling behavior, and seasonal activity; environmental conditions (vegetation, wind, humidity, rainfall) bias sound transmission and detection; deployments face battery, storage, and equipment-failure constraints; and individual sensors give limited information on caller distance or number, complicating abundance estimates.
- Stated in the paper: acoustic indices correlate only moderately with biodiversity metrics and are sensitive to noise, weather, and recording-device differences; ML models depend on unevenly distributed labeled data, may not generalize across habitats or devices, and often operate as black boxes.
- Evident from the text: as a narrative review it does not report a systematic search or inclusion protocol, and several forward-looking claims (global listening networks, ecosystem digital twins, "animal Google Translate") are presented as visions or early-stage work rather than demonstrated outcomes.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
26 Valid · 2 Not valid · 91 Not checked · 38 Own results and interpretations
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“Many signal-processing features are also sensitive to background noise, weather conditions, and anthropogenic sounds, which can distort ecological interpretations”
What the sources say
- Supportedcites Quinn et al. 2023 matched to “Soundscape components inform acoustic index patterns and refine estimates of bird species richness” (2023) source
The Introduction says that anthropogenic sounds, weather sounds, and ambient noise complicate the use of acoustic indices as ecosystem-monitoring tools. Discussion §4.3 adds that other sound sources significantly affect the indices and make their ecological interpretation non-trivial. (LLM, conf=0.93)
- Not checked yetcites Zhang et al. 2024 matched to “Influence of recording devices and environmental noise on acoustic index scores: Implications for bird sound-based assessments” (2024) source
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Neuromorphic systems—modeled after the spiking logic of biological cochleae—may offer substantial improvements for fieldwork. They deliver orders-of-magnitude energy savings while keeping classification accuracy impressively high”
What the sources say
- Partly supportedcites Lenk et al. 2023 matched to “Neuromorphic acoustic sensing using an adaptive microelectromechanical cochlea with integrated feedback” (2023) sourceExperimental — accuracy not yet measured
The Sensing properties section describes an adaptive bio-inspired cochlea and says improved energy efficiency is expected; the Dynamical adaptation section reports a spike-like sensor output. But the paper does not establish orders-of-magnitude energy savings, high classification accuracy, or improvements in fieldwork. Its Conclusions present the device as a potential input stage for speech processing, not a demonstrated classification system. (LLM, conf=0.87)
- Not checked yetcites Wu et al. 2019
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“Together, these indices are increasingly used as quick, cost-effective barometers of ecosystem state, now deployed everywhere from single-species surveys to continent-wide biodiversity monitoring”
What the sources say
- Supportedcites Bradfer-Lawrence et al. 2023 matched to “Using acoustic indices in ecology: Guidance on study design, analyses and interpretation” (2023) source
The Abstract (p. 1-2) states that acoustic indices are 'currently one of the most widely applied tools in ecoacoustics', 'relatively straightforward and fast to calculate', and that they 'can offer insights into the state of organisms, populations, and ecosystems ... for biodiversity monitoring, conservation and future research' — supporting the claim's 'increasingly used', 'quick, cost-effective barometers of ecosystem state'. The span of deployment is covered too: the Introduction (Section 1) frames ecoacoustics as working 'across levels from individual organisms to communities and landscapes' with 'low-cost, robust, passive acoustic recorders', and Section 2 offers recommendations 'for the use of acoustic indices in the study of both whole soundscapes and individual taxa and species', including 'detection of large-scale soundscape patterns'. Only the literal phrase 'continent-wide' is not used, but the source clearly treats the method as applied from single-species to landscape/regional biodiversity monitoring, so this is a broad background statement rather than an overstatement. (LLM, conf=0.82)
- Not checked yetcites Gage et al. 2001 matched to “Assessment of ecosystem biodiversity by acoustic diversity indices” (2001) source
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Environmental conditions further shape what sensors record: vegetation structure, wind, humidity, and rainfall influence sound transmission and detection distance, potentially introducing spatial and temporal biases across large monitoring networks”
What the sources say
- Partly supportedcites Winiarska et al. 2024 matched to “Detection ranges of forest bird vocalisations: guidelines for passive acoustic monitoring” (2024) sourceExperimental — accuracy not yet measured
The Results (Detection ranges: general overview; no page supplied) report that forest type, humidity, wind, and season affected detection distance or probability, supporting the possibility of spatial and temporal differences in recordings. But the experiment took place without rain (Materials and methods, Design of transmission experiment), so it does not establish a rainfall effect. It also did not study biases across large monitoring networks. (LLM, conf=0.88)
- Partly supportedcites Yip et al. 2026 matched to “Contributions of environmental conditions and sound characteristics to differences in perceptibility: Recommendations for passive acoustic monitoring” (2026) sourceExperimental — accuracy not yet measured
The Results section (“Effects of Song Structure, Vegetation, and Environmental Conditions,” page unavailable) reports that vegetation type, wind, and humidity affect perceptibility; the Introduction says vegetation- and weather-related variation can change the area surveyed even at the same location over time. The Introduction also discusses potential systematic bias as monitoring expands in space and time. Rainfall, however, is not identified among the conditions studied or discussed in the passages found, so that specific part of the claim is unsupported. (LLM, conf=0.88)
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Differences in microphones, recording settings, and sampling rates can further affect spectral features and index values, complicating comparisons across studies”
What the sources say
- Supportedcites Luna-Naranjo et al. 2023 matched to “Quantifying and mitigating recorder-induced variability in ecological acoustic indices” (2023) source
The Conclusion identifies sampling frequency, microphone frequency response, and gain settings as sources of recorder-induced variability in ecological acoustic indices. Sections 4.1 and 4.2 describe effects on index values, while the Introduction says device differences prevent direct comparison of acoustic features. Together, these support the claim about spectral features and comparisons across studies. (LLM, conf=0.93)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
63 more claims · 11 with their cited sources checked
Create a free account to see all 157 claims and to verify the 150 references still open.
- It may state two separate claims as one. 66
- Probably not a substantial claim. 38
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Introduction
“(Growcott et al. 2024; Pedroza et al. 2020)”
§ References
“Growcott J, Lobora A, Markham A, Searle CE, Wahlström J, Wijers M, Simmons BI (2025) The secret acoustic world of leopards: a paired camera trap and bioacoustics survey facilitates individual identification of leopards via their roars.”
Citations
Cites
1 retracted
Items 201–216 of 216
The sonic heritage of ecosystems
Monacchi, D. (2017) · doi:10.18146/soima2015.3.12 · Cited by 3 · cited 1×
“…such as Colombia’s Environmental Sound Collection (Mendoza-Henao et al. 2023) and Biosphere Soundscapes (Barclay and Gifford 2018) already show how systematic documentation—even “paleosoundscapes” from pristine habitats can become cultural, scientific, and conservation resources (Monacchi 2017).”
Acoustic ecology in UNESCO biosphere reserves
Barclay, L., & Gifford, T. (2017) · VIUSpace (Vancouver Island University Library) · doi:10.25316/ir-386 · Cited by 3 · cited 2×
“Proj-ects such as Colombia’s Environmental Sound Collection (Mendoza-Henao et al. 2023) and Biosphere Soundscapes (Barclay and Gifford 2018) already show how systematic documentation—even “paleosoundscapes” from pristine habitats can become cultural, scientific, and conservation resources (Monacchi…”
Gupta, A. S., Kubicek, B., McCarthy, R. A., Linhardt, T., Hermann, L., & Kemerling, M. (2021) · The Journal of the Acoustical Society of America · doi:10.1121/10.0004451 · Cited by 2 · cited 2×
“…archiving, and querying of large acoustic datasets (Aide et al. 2013; K. F. Darras et al. 2024; Martínez Balvanera et al. 2025 (Becker et al. 2018; Gupta et al. 2021); autoencoder-based clustering of vocalizations (Bergler et al. 2019); hierarchical Dirichlet process HMMs for acoustic unit…”
Behavior Classification of Egyptian Fruit Bat (Rousettus aegyptiacus) From Calls With Deep Learning
Yılmaz, B., Sen, M., Maşazade, E., & Beşkardeş, V. (2022) · Advances in computational intelligence and robotics book series · doi:10.4018/978-1-7998-8686-0.ch004 · Cited by 2 · cited 1×
“…with high precision by pairing accelerometer data with sound (Wijers et al. 2018), and deep networks often outperform classic approaches at teasing apart the social contexts of Egyptian fruit bat calls, reliably separating vocalizations linked to biting, feeding, and fighting (Yılmaz et al. 2022).”
Network Digital Twins for Multi-Domain Planning and Operations
Goudy, M. (2024) · doi:10.1109/comcas58210.2024.10666171 · Cited by 2 · cited 1×
“Insights from the digital-twin literature suggest how real-time sensor data could feed into predictive simulators (Goudy 2024).”
Augustin, J., Zsebők, S., Kovács, D., Jánki, Z., Bánhalmi, A., Soltész, Z., Seffer, P., Bilicki, V., & Garamszegi, L. Z. (2026) · PLOS One · doi:10.1371/journal.pone.0343060 · Cited by 2 · cited 1×
“For example, mosquito wingbeat frequencies—often used for AI-based species recognition—are strongly influenced by factors such as temperature, sex, and body size, which can reduce classification reliability if such variability is not represented in training data (Augustin et al. 2026).”
Self-citationDisplay of Sound Spectrographs in Real Time
Harris, C. M., & Waite, W. M. (1963) · The Journal of the Acoustical Society of America · doi:10.1121/1.1918601 · Cited by 1 · cited 1×
“By the 1960s, electronics had advanced enough to allow real-time spectrogram displays on oscilloscopes (Harris and Waite 1963).”
From visible speech to voiceprints - blessing and curse of sound spectrography
Braun, A. (2017) · doi:10.21437/hscr.2017-8 · Cited by 1 · cited 1×
“For the first time, researchers could produce continuous spectrograms—plot-ting frequency against time—and measure communication signals quantitatively (Braun 2017; Koenig et al. 1946).”
Wall, C. C., Anderson, C., & Martinez, V. (2021) · The Journal of the Acoustical Society of America · doi:10.1121/10.0004521 · Cited by 1 · cited 2×
“…schemes now cut file sizes by up to 85% while preserving the spectral fidelity needed for research (Diego-Tortosa et al. 2025; S. B. Martin et al. 2021), and standardized metadata systems are fast becoming essential for reproducibility and cross-project collaboration (Wall et al. 2021a, b ).”
Parselmouth for bioacoustics: Integrating Praat into the Python scientific ecosystem
Jadoul, Y., Duengen, D., & Ravignani, A. (2022) · The Journal of the Acoustical Society of America · doi:10.1121/10.0010550 · Cited by 1 · cited 1×
“Python libraries like OpenSound-scape, librosa, Parselmouth and scikit-maad now let anyone build modular detection and classification pipelines (Jadoul et al. 2022; Lapp et al. 2023; McFee et al. 2015; Ulloa et al. 2021).”
The Ethics of Ownership: Developing Guidelines on Data Ownership in Participatory Researchers
Smith, S. (2024) · Septentrio Conference Series · doi:10.7557/5.7799 · Cited by 1 · cited 1×
“Consumer devices also contribute variable-quality recordings, introducing noise literally and figuratively but governance models such as fractional ownership schemes and creative attribution systems are emerging to keep contributors acknowledged and engaged (Smith 2024; Trejo et al. 2021).”
A generic AI pipeline for identifying known species and discovering new species
Campos‐Cerqueira, M. (2023) · The Journal of the Acoustical Society of America · doi:10.1121/10.0018601 · Cited by 1 · cited 1×
“These approaches point toward a future where we can estimate biodiversity baselines directly from raw sound-scapes—no exhaustive hand-labeling needed (Briggs et al. 2016; Campos-Cerqueira 2023; Cuevas et al. 2017; Rama Rao et al. 2018).”
Feature interpretation in underwater acoustics
Potty, G. R. (2024) · The Journal of the Acoustical Society of America · doi:10.1121/10.0035276 · Cited by 1 · cited 1×
“In addition, deep-learning approaches often operate as black boxes, making it difficult to interpret the acoustic features driving classification decisions (Heinrich et al. 2024; Potty 2024).”
Bridge, M. C. (2025) · Historical Studies in the Natural Sciences · doi:10.1525/hsns.2025.55.3.209 · Cited by 1
Quantum-Inspired Genetic Algorithm for Robust Source Separation in Smart City AcousticsPreprint
Quan, M. K., Wijayasundara, M., Setunge, S., et al. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2504.07345 · Cited by 1
The Next Wave of Passive Acoustic Data Management: How Centralized Access Can Enhance Science
Wall, C. C., Haver, S. M., Hatch, L. T., Miksis-Olds, J. L., Bochenek, R., Dziak, R. P., & Gedamke, J. (2021) · University of New Hampshire Scholars Repository (University of New Hampshire at Manchester) · doi:10.3389/fmars.2021.703682 · cited 2×
“…schemes now cut file sizes by up to 85% while preserving the spectral fidelity needed for research (Diego-Tortosa et al. 2025; S. B. Martin et al. 2021), and standardized metadata systems are fast becoming essential for reproducibility and cross-project collaboration (Wall et al. 2021a, b ).”
Research works printed but not linked: 6
- Becker S, Ackermann M, Lapuschkin S, Müller K, Samek W (2018) Interpreting and explaining deep neural networks for classification of audio signals. arXiv.Org
- Jančovič P, Köküer M, Zakeri M, Russell M (2013) Unsupervised discovery of acoustic patterns in bird vocalisations employing DTW and clustering. In: European Signal Processing Conference
- Kamau E, Fedder B, Winter G (2010) The Nagoya protocol on access to genetic resources and benefit sharing: What is new and what are the implications for provider and user countries and the scientific community? Law Environ Dev J 6(3):246–262
- Leso V, Fontana L, Finiello F, De Cicco L, Ercolano ML, Iavicoli I (2020) Noise induced epigenetic effects: a systematic review. Noise Health 22(107):77–89
- Ropert-Coudert Y, Wilson RP (2005) Trends and perspectives in animal-attached remote sensing. Front Ecol Environ 3(8):437–444. https://doi.org/10.1890/1540-9295 · doi:10.1890/1540-9295
- Wu J, Pan Z, Zhang M, Das RK, Chua Y, Li H (2019) Robust sound recognition: a neuromorphic approach. Interspeech
Metadata
- Authors
- Zsebők, S.
- Affiliations
- Eötvös Loránd University (HU)
- Corresponding author
- Zsebők, Sándor
- ORCID
- Zsebők, Sándor
- Venue
- Biologia Futura
- Published
- 2026-06-09
- Volume
- 77
- Issue
- 3
- Pages
- 425-441
- ISSN
- 2676-8607 (ISSN-L)
- 2676-8615
- OpenAlex
- W7164031808
- Cited by
- 0
- References
- 211
- FWCI
- 0.00
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- Biologia Futura (publishedVersion, cc-by)
- PubMed (publishedVersion)
- Funders
- HUN-REN Centre for Ecological Research
- Topics
- Life Sciences › Agricultural and Biological Sciences › Ecology, Evolution, Behavior and Systematics › Animal Vocal Communication and Behavior
- Physical Sciences › Environmental Science › Ecology › Marine animal studies overview
- Physical Sciences › Environmental Science › Ecology › Wildlife Ecology and Conservation
- Keywords
- bioacoustics, animal communication, individual identification, cultural evolution, acoustic indices, biodiversity monitoring, machine learning, transfer learning, explainable AI, autonomous recording units, edge computing, data governance, acoustic privacy
- MeSH terms
- Machine Learning, Acoustics, Animals, Conservation of Natural Resources, Ecology, Biological Evolution, Humans, Ecosystem
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-29, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-09-29 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.