Multiple taxa: common nightingale and bats / Mobile bioacoustics methods comparison

Phones can record wildlife, but they are not interchangeable sensors

Primary studies reviewed 2021–2024 · Note reviewed 2026-10-13

Two different studies show that smartphones can support bounded wildlife-acoustic tasks, but they do not create a uniform sensor. Jäckel et al. compared 5,679 Berlin citizen-science nightingale recordings with expert recordings: 53% of citizen recordings were categorized as complete songs versus 100% of expert recordings, and signal-to-noise ratios were lower for citizen recordings. In a small controlled playback comparison, estimated frequency limits varied by device. The authors nevertheless considered the citizen recordings useful for questions based on relative song-type occurrence, while cautioning about exact acoustic measurements. Gili et al. tested built-in microphones for selected 8–24 kHz bat calls. Their field/pilot work covered 35 phone models and nine bat species, but found device-model variation, occasional aliasing and loss of calls near the upper detection limit; the authors recommend preliminary detectability checks and expert verification of at least some identifications.

What we reviewed · Selected full-text methods, results, discussion, data-availability and rights review

For Jäckel et al. (2021), reviewed the PLOS paper's methods, citizen-recording quality results, device playback comparison, discussion, rights and Zenodo data statement; the cited Zenodo deposit remains unreviewed. For Gili et al. (2024), reviewed the full publisher article and the March 13, 2024 Dryad record, visible file list and embedded README structure, as well as Dryad's CC0 and responsible-sharing policies. The separate README download returned HTTP 403. No audio archive, article supplement or code was acquired or reproduced, and no independent specialist review was completed.

No independent bioacoustics or mobile-audio specialist has reviewed this Talk2Nature synthesis or proposed device protocol.

Study at a glance

Nightingale study: Citizen scientists used the Naturblick app; 5,679 recordings from Berlin in 2019 were compared with expert recordings. Recording instructions were intentionally limited, so the sample also reflects varied user behavior and context.
Nightingale quality result: 53% of citizen-science recordings versus 100% of expert recordings were classified as complete songs. Citizen recordings had lower signal-to-noise ratios; frequency estimates varied among devices in the small playback comparison.
Bat device comparison: Selected built-in phone microphones were compared with a reference bat detector for calls in an 8–24 kHz range. The paper reports model-level differences and calls near the upper limit were more likely to become undetectable with distance.
Bat participation pilot: The pilot tested 35 phone models across 80 volunteer sessions and reported nine recorded species; some recordings remained unidentified and five tested models showed aliasing artifacts. The authors recommend preliminary device checks and expert validation.
Dryad deposit inspection: The versioned Dryad page lists an Appendix S6 examples archive (17.46 MB), a full mobile-recordings archive (3.89 GB), and a 6.49 KB README. Its embedded README describes 80 sessions with device model, operating system, date, species folders and site-code-to-municipality mappings. Dryad's current guidance says hosted datasets are CC0 1.0; the item page itself shows no record-specific license. The README download returned HTTP 403 and no audio was acquired.
Data-use boundary: CC0 per Dryad policy supports reuse, but metadata review does not establish archive contents, label quality, location masking, consent scope or benchmark fitness. Keep the record metadata-only; if justified, inspect the smaller examples archive first under a documented protocol.
Product decision: Make phone variability visible before recording. The existing eight-second check measures temporary browser input only; it does not calibrate microphones or establish species coverage. Preserve a simple local observation path while reserving comparative scientific claims for a reviewed device protocol.

Reading location: Jäckel et al. 2021, PLOS ONE 16(6):e0253763: Materials and methods; citizen-science recording-quality results; device playback test; Discussion; data availability and rights. Gili et al. 2024, Biodiversity and Conservation 33:1623–1643: mobile-device protocol; results; Discussion; data availability and rights. Checked 2026-10-13.

What this does — and doesn’t — establish

These studies cover different species, frequency ranges, devices, apps, protocols and outcomes; their results cannot be pooled into a general phone-quality score. The nightingale work is a Berlin project and its controlled device set was small. The bat work focuses on selected calls within 8–24 kHz at Italian and Spanish sites; it does not establish general ultrasonic capture above that range. Neither study validates current phones, current operating systems, browser capture or Talk2Nature's exact audio path. Acoustic recording or species-level detection does not demonstrate receiver comprehension or call meaning.

What we can build on

Keep the phone as an accessible observation tool, but treat device and capture path as part of the measurement. A browser level check is not microphone calibration, and a WAV export does not make recordings comparable across phones. Any later scientific protocol should test the chosen phone models and app path against a reference recorder for the target species and frequency range, then preserve the untouched source and protocol metadata. Collect only the minimum device information needed, with consent; do not fingerprint participants. Keep detection, identification, behavior association and meaning as separate claims.

OPEN METHODS QUESTION

Independent review of smartphone recording checks

We invite bioacoustics, acoustic-measurement or mobile-audio specialists to assess whether a small, no-animal reference test of device and app-path variation would answer a useful question without overstating microphone calibration or field suitability. This is a critique of a possible next step; no test protocol, pass/fail threshold or device recommendation has been approved.

Would a no-animal reference test answer a useful prerequisite question for a defined acoustic task, or is it too detached from field conditions to guide the app?
Which capture paths should be compared first, and what source/reference recording would separate microphone, operating-system/browser processing and app settings?
What minimum device/session metadata supports reproduction without collecting serial numbers, account identifiers, precise locations or persistent device fingerprints?
Which measurements and uncertainty summaries matter for a selected frequency range, and is any pass/fail threshold defensible?
What can the app's short input-level check honestly tell a visitor, and what result would change the decision to build a more advanced device-check workflow?

Review status: No independent specialist review has been received; no bench protocol, pass/fail threshold or device recommendation is approved, and no physical-device test is claimed.

Read the focused reviewer brief ↗Submit a public methods review ↗

Do not attach recordings, private research material, personal information or sensitive locations. A review is not an endorsement or partnership.

The next reading task

Ask an independent bioacoustics/mobile-audio reviewer to define a small, no-animal device-characterization protocol for the exact Android and iOS capture paths: paired phone/reference recordings, target frequency ranges, browser audio processing, device/session grouping, and pass/fail limits. For the Dryad bat deposit, review the 17.46 MB Appendix S6 examples only if a concrete research question justifies it, with a recorded checksum and file-level privacy/location screen; defer the 3.89 GB archive. The linked Zenodo nightingale deposit remains unreviewed. Keep participant intake closed until protocol, consent, privacy, rights and withdrawal gates are approved.

Go to the sources

Citizen-science and expert nightingale recordings ↗
Jäckel et al. (2021), PLOS ONE 16:e0253763. A Berlin citizen-science study compared app recordings and professional recordings for nightingale song; it also tested device-specific acoustic deviations.
Built-in mobile microphones for bat monitoring ↗
Gili, Bertolino & Rolando (2024), Biodiversity and Conservation 33:1623–1643. Field and participatory comparisons of built-in mobile microphones with a bat detector across selected low-frequency bat calls.
Bat recordings collected by mobile devices ↗
Dryad DOI 10.5061/dryad.2280gb5z4, cited by the 2024 bat-monitoring paper as the location of mobile-device recordings.
Dryad data reuse and CC0 policy ↗
Dryad's current reuse guide states that all datasets hosted by Dryad are published under CC0 1.0 and recommends scholarly citation. This platform policy does not replace inspection of a specific record's files, provenance, privacy or sensitive-species context.
Dryad responsible data-sharing guidance ↗
Dryad's data-sharing guidance says to anonymize human-subject data and consider masking locations for vulnerable or endangered species; it also recommends complete READMEs and reproducible context.

Original editorial note by Talk2Nature contributors. AI-assisted, source-grounded synthesis; not a peer review or an independent replication. Linked works retain their own rights.