AircraftWitness: How Often Aircraft Can Locate Recordings

Kris Atallah

AI summary · arXiv: 2611.08791v1 · cs.SD

AircraftWitness: How Often Aircraft Can Locate Recordings

Kris Atallah (New York University, New York, USA)

Original paper Submitted to arXiv on 05 Nov. 2026 16 pages, 8 figures, 6 tables. Code and dataset: https://github.com/nyu-kr/aircraftwitness License: CC0 1.0 Also in: fr

How to read this summary? Illustrated guide

The essentials

AircraftWitness finds that aircraft audio can locate some outdoor recordings, but successful localization is uncommon and depends on clear passes and accurate timing.

Based on the abstract only: check the details in the original paper.

Question

The paper asks how often aircraft sounds, matched with public ADS-B flight records, can reveal the location of outdoor audio recordings. It tests localization under blind conditions and compares performance with chance.

Method

The main preregistered test covered 24 new locations, using recorder clocks, 300 seconds of timing slack, and a 20 km square. The authors also evaluated three locators on 360 clips from seven GPS-timed locations and on 1,600 randomly selected public recordings; they released a benchmark and pass labels for 11.7 million recordings.

Findings

Result The best of three locators placed clear-pass clips within 1 km at four of 24 locations, compared with 0.57 expected from a uniform guess; the reported corrected p-value was 0.0065.
See the passage in the paper
The best of three locators places clear-pass clips within 1 km at 4 places, where a uniform guess predicts 0.57 (p = 0.0065 after correcting for the pick).
Result When the time was wrong or audio was swapped, the best locator did not place any recordings within 1 km.
See the passage in the paper
With the wrong time or swapped audio it places none.
Result On 360 GPS-timed clips, two locators placed 6% and 8% within 1 km, and a unit selected after the run located recordings at four and five of the seven places.
See the passage in the paper
On 360 clips from 7 places of a GPS-timed network, two locators place 6% and 8% within 1 km. By place, a unit chosen after that run, they hit 4 and 5 of 7 where a uniform guess predicts 2.3.
Result On randomly selected public recordings, the locators returned few answers and placed at most two recordings.
See the passage in the paper
On 1,600 random public recordings, picked and cut without the truth in a time-bounded run, the three locators answer at most 20 times and place at most 2.

Limitations

The blind public-recording test produced few answers, and at most two recordings were placed.
See the passage in the paper
On 1,600 random public recordings, picked and cut without the truth in a time-bounded run, the three locators answer at most 20 times and place at most 2.
The main test covers only 24 new places, so the reported performance may not represent other recording locations.
See the passage in the paper
It covers 24 new places, with the recorders' own clocks, a 300 s slack and a 20 km square.

Why it matters

The authors present AircraftWitness as a benchmark and release pass labels for 11.7 million recordings, enabling further evaluation of aircraft-based audio localization. Their results show the approach can outperform chance in some controlled cases, while successful localization of ordinary public recordings remains limited.

Written by AI from the paper. Passages are checked word for word against the text we analysed; the rest may contain errors.

AI-generated keywords: Aircraft localization ADS-B acoustic analysis Doppler patterns outdoor recordings

Read further

Comprehensive Summary
Layman's Summary
Blog article
Original abstract

The paper investigates how often aircraft activity can help identify where an outdoor audio recording was made. Its premise is that propellers and rotors can produce distinctive Doppler patterns, while public ADS-B archives contain aircraft positions and flight tracks. The authors test whether matching audio patterns to logged aircraft movements can locate recordings and compare performance with chance. The main test was registered and timestamped before the audio was examined. The authors report that both amendments to the test were made before any results were reviewed. The test used recordings from 24 new locations, the recorders’ own clocks, a 300-second timing allowance, and a 20-by-20-kilometre search area. For clips with clear aircraft passes, the best-performing locator placed recordings within one kilometre at four locations. Under the stated uniform-guess baseline, success was expected at 0.57 locations. The reported significance was p = 0.0065 after correcting for selection of the best of three locators. The system produced no successful placements when the time was incorrect or the audio was swapped. The other two locators succeeded at one and two locations, respectively. A second evaluation used 360 clips from seven locations in a network of GPS-timed recorders. Two locators placed 6% and 8% of clips within one kilometre. In a location-level analysis, a locator selected after the evaluation placed recordings successfully at four and five of the seven locations; the uniform-guess baseline predicted success at 2.3 locations. The abstract does not explain how clips were selected or how location-level success was defined. The third evaluation tested 1,600 randomly selected public recordings. The recordings were selected and cut without access to their true locations, in a time-bounded run. Across the three locators, no more than 20 recordings received a location estimate, and no more than two were placed successfully. This contrasts with the more promising results for clear-pass clips and the GPS-timed network. It suggests that practical yield depends strongly on recording conditions and the evaluation setting. The authors report releasing AircraftWitness, a benchmark, and pass labels for 11.7 million recordings. Overall, the abstract presents evidence that aircraft audio can sometimes help locate recordings, while showing that successful localization was uncommon in a large, randomly selected public dataset. The wrong-time and swapped-audio checks indicate that the main-test successes did not recur under those altered conditions. This summary is based only on the title and abstract; the full text was unavailable. A fuller assessment would require details that the abstract does not provide, including the locator algorithms, the definition of a “clear pass,” uncertainty estimates, dataset composition, and how the chance baselines were implemented. The abstract also does not establish how well the findings generalize beyond the reported datasets and settings.

Can an Aircraft Reveal Where a Recording Was Made?

In “AircraftWitness: How Often Aircraft Can Locate Recordings,” Kris Atallah investigates whether sounds from passing aircraft can help identify where an outdoor recording was captured. The idea is simple but intriguing: propellers and rotors can create distinctive changes in pitch as an aircraft moves, while public ADS-B archives document aircraft positions and flight paths. If an audio pattern can be matched to a logged flight, the aircraft’s route might offer clues to the recorder’s location.

The study asks not only whether this approach can work, but how often it succeeds—and whether its performance is better than chance. Its results offer encouraging evidence under some conditions, alongside an important caution: success was uncommon in a large test using randomly selected public recordings.

A Test Planned Before the Audio Was Examined

One strength of the main evaluation is that its test was registered and timestamped before the researchers examined the audio. The abstract also says that both amendments to the test were made before results were reviewed. This ordering helps guard against changing the evaluation after seeing which methods or recordings performed well.

The main test used recordings from 24 new locations. It relied on the recorders’ own clocks, allowed a 300-second timing difference, and searched within a 20-by-20-kilometre area. For recordings described as having clear aircraft passes, the best-performing of three location-finding methods placed recordings within one kilometre at four locations. The study’s stated uniform-guess baseline predicted success at 0.57 locations.

After correcting for the selection of the best-performing method out of three, the reported significance was p = 0.0065. The other two methods succeeded at one and two locations, respectively. The system did not produce successful placements when the time was wrong or when audio was swapped. Together, these checks indicate that the main-test successes did not simply recur under those altered conditions.

Results Depend on the Evaluation Setting

GPS-timed recorder network

A second evaluation used 360 clips from seven locations in a network of GPS-timed recorders. Two methods placed 6% and 8% of clips within one kilometre. The study also reports a location-level analysis: a method selected after the evaluation placed recordings successfully at four and five of the seven locations, compared with 2.3 locations predicted by the uniform-guess baseline.

Those location-level figures sound promising, but the abstract does not explain how clips were selected or exactly how success at a location was defined. Without those details, it is difficult to interpret how the clip-level and location-level results relate to one another, or how broadly the findings apply.

Randomly selected public recordings

The third evaluation gives a more sobering picture of practical performance. It tested 1,600 randomly selected public recordings. The recordings were selected and cut without access to their true locations, in a time-bounded run. Across all three methods, no more than 20 recordings received a location estimate, and no more than two were placed successfully.

This contrast matters. Aircraft sounds may be useful when a recording contains a clear pass or is collected in a carefully timed network, yet the approach appears to have a low yield in this large, randomly selected set. Performance therefore seems closely tied to recording conditions and evaluation design—not merely to whether aircraft can, in principle, be tracked.

What the Findings Do—and Do Not—Show

The results support a measured conclusion: aircraft audio can sometimes help locate an outdoor recording, and the main test performed better than its stated chance baseline. But the public-recording evaluation shows that successful localization was uncommon in a broader sample. The wrong-time and swapped-audio checks add useful evidence that the main-test successes depended on the intended timing and audio pairing.

The abstract also reports the release of AircraftWitness, a benchmark, and pass labels for 11.7 million recordings. These resources could help others investigate the method and compare future approaches.

Still, a fuller assessment would require information not included in the abstract: how the location-finding methods work, what qualifies as a “clear pass,” how uncertainty was measured, what the datasets contain, and how the chance baselines were implemented. The abstract also does not establish whether the results generalize beyond the datasets and settings described.

A Promising Clue, Not a Universal Locator

AircraftWitness presents aircraft tracking as a possible source of location clues in outdoor audio, rather than a dependable solution for every recording. Its strongest results came from selected or specially structured evaluations; its largest random test yielded few successful placements. That balance makes the paper’s central message appropriately cautious: aircraft can sometimes help locate a recording, but how often they help depends substantially on the data and conditions.

Created on 09 Nov. 2026

Was this synthesis useful?

Score: 0

Ask questions about this paper to our AI assistant

How to chat with this paper? Illustrated guide

Compare with the PDF

You can also chat with multiple papers at once here.

AI assistant instructions?

Tools for this paper

Explain it differently 1 free rewrite left today

Save to a collection Group papers in collections to compare them, chat with them and export them. Continue with Google Create a free account Log in

Export BibTeX · RIS (Zotero, Mendeley, EndNote) · Obsidian (Markdown) · Notion (Markdown)

Advanced tools Premium

For researchers who follow a field closely: citation alerts, journal-club slides, version comparison and Zotero export. Discover Premium

Zotero

Citation alerts Citation alerts guide

Who works on it? The people and institutions who publish on its topic

Weekly letter Presented in "Computing beyond AI this week" of 9 November 2026. Computing beyond AI this week: What noise reveals, and what guarantees conceal

Journal club Premium+ Slides to present it Propose for the journal club

Versions What changed between versions

Talk about it in public Premium+ Press kit: release, questions, hype check Press kit guide

Papers around this one

See them on a map

Ranked by how close their title and abstract are to this paper’s. Select some to read them together.

Closest papers

Related Broader context
A Search for Technosignatures Around 11,680 Stars with the Green Bank Telescope at 1.15-1.73 GHz
Jean-Luc Margot, Megan G. Li, Pavlo Pinchuk et al. · 2023 · AJ 166 206 (2023) · cited 15 times · AI summary available

The Green Bank Telescope search identifies narrowband radio technosignatures, whereas AircraftWitness uses aircraft sounds and flight records to locate outdoor recordings.

Abstract
We conducted a search for narrowband radio signals over four observing sessions in 2020-2023 with the L-band receiver (1.15-1.73 GHz) of the 100 m diameter Green Bank Telescope. We pointed the telescope in the directions of 62 TESS Objects of Interest, capturing radio emissions from a total of ~11,860 stars and planetary systems in the ~9 arcminute beam of the telescope. All detections were either automatically rejected or visually inspected and confirmed to be of anthropogenic nature. In this work, we also quantified the end-to-end efficiency of radio SETI pipelines with a signal injection and recovery analysis. The UCLA SETI pipeline recovers 94.0% of the injected signals over the usable f…
Related Broader context
UAV's Rotor Micro-Doppler Feature Extraction Using Integrated Sensing and Communication Signal: Algorithm Design and Testbed Evaluation
Jiachen Wei, Dingyou Ma, Feiyang He et al. · 2024 · Signal Processing · cited 24 times · Not summarized yet

The UAV study extracts rotor micro-Doppler from sensing signals to detect hovering drones, while AircraftWitness uses aircraft Doppler patterns in audio for location.

Abstract
With the rapid application of unmanned aerial vehicles (UAVs) in urban areas, the identification and tracking of hovering UAVs have become critical challenges, significantly impacting the safety of aircraft take-off and landing operations. As a promising technology for 6G mobile systems, integrated sensing and communication (ISAC) can be used to detect high-mobility UAVs with a low deployment cost. The micro-Doppler signals from UAV rotors can be leveraged to address the detection of low-mobility and hovering UAVs using ISAC signals. However, determining whether the frame structure of the ISAC system can be used to identify UAVs, and how to accurately capture the weak rotor micro-Doppler sig…
Related Broader context
A deep-learning search for technosignatures of 820 nearby stars
Peter Xiangyuan Ma, Cherry Ng, Leandro Rizk et al. · 2023 · Instrumentation and Methods for Astrophysics · cited 41 times · AI summary available

The deep-learning search looks for Doppler-drifting radio technosignatures, while AircraftWitness analyzes aircraft Doppler curves in audio to infer recording locations.

Abstract
The goal of the Search for Extraterrestrial Intelligence (SETI) is to quantify the prevalence of technological life beyond Earth via their "technosignatures". One theorized technosignature is narrowband Doppler drifting radio signals. The principal challenge in conducting SETI in the radio domain is developing a generalized technique to reject human radio frequency interference (RFI). Here, we present the most comprehensive deep-learning based technosignature search to date, returning 8 promising ETI signals of interest for re-observation as part of the Breakthrough Listen initiative. The search comprises 820 unique targets observed with the Robert C. Byrd Green Bank Telescope, totaling over…
Related Broader context
Privacy-Preserving by Design: Indoor Positioning System Using Wi-Fi Passive TDOA
Mohamed Mohsen, Hamada Rizk, Moustafa Youssef · 2023 · Artificial Intelligence · cited 10 times · AI summary available

PassiFi estimates indoor positions from passive Wi-Fi TDoA measurements; AircraftWitness estimates outdoor recording locations from aircraft audio and ADS-B records.

Abstract
Indoor localization systems have become increasingly important in a wide range of applications, including industry, security, logistics, and emergency services. However, the growing demand for accurate localization has heightened concerns over privacy, as many localization systems rely on active signals that can be misused by an adversary to track users' movements or manipulate their measurements. This paper presents PassiFi, a novel passive Wi-Fi time-based indoor localization system that effectively balances accuracy and privacy. PassiFi uses a passive WiFi Time Difference of Arrival (TDoA) approach that ensures users' privacy and safeguards the integrity of their measurement data while st…
Related Broader context
Retrieval in Long Surveillance Videos using User Described Motion and Object Attributes
Greg Castanon, Mohamed Elgharib, Venkatesh Saligrama et al. · 2014 · Computer Vision and Pattern Recognition · cited 24 times · Not summarized yet

The surveillance-video method retrieves segments using motion and object attributes, whereas AircraftWitness matches aircraft sounds to flight records to locate recordings.

Abstract
We present a content-based retrieval method for long surveillance videos both for wide-area (Airborne) as well as near-field imagery (CCTV). Our goal is to retrieve video segments, with a focus on detecting objects moving on routes, that match user-defined events of interest. The sheer size and remote locations where surveillance videos are acquired, necessitates highly compressed representations that are also meaningful for supporting user-defined queries. To address these challenges we archive long-surveillance video through lightweight processing based on low-level local spatio-temporal extraction of motion and object features. These are then hashed into an inverted index using locality-s…
Related Broader context
Drone navigation and license place detection for vehicle location in indoor spaces
Moa Arvidsson, Sithichot Sawirot, Cristofer Englund et al. · 2023 · Computer Vision and Pattern Recognition · cited 2 times · AI summary available

The drone system maps parked vehicles from detected license plates, while AircraftWitness locates outdoor audio recordings using aircraft sounds and ADS-B data.

Abstract
Millions of vehicles are transported every year, tightly parked in vessels or boats. To reduce the risks of associated safety issues like fires, knowing the location of vehicles is essential, since different vehicles may need different mitigation measures, e.g. electric cars. This work is aimed at creating a solution based on a nano-drone that navigates across rows of parked vehicles and detects their license plates. We do so via a wall-following algorithm, and a CNN trained to detect license plates. All computations are done in real-time on the drone, which just sends position and detected images that allow the creation of a 2D map with the position of the plates. Our solution is capable of…
Related Broader context
GPS-Spoofing Attack Detection Mechanism for UAV Swarms
Pavlo Mykytyn, Marcin Brzozowski, Zoya Dyka et al. · 2023 · Cryptography and Security · cited 28 times · AI summary available

The UAV study detects GPS spoofing in drone swarms, while AircraftWitness uses aircraft audio and public flight records to estimate where recordings were made.

Abstract
Recently autonomous and semi-autonomous Unmanned Aerial Vehicle (UAV) swarms started to receive a lot of research interest and demand from various civil application fields. However, for successful mission execution, UAV swarms require Global navigation satellite system signals and in particular, Global Positioning System (GPS) signals for navigation. Unfortunately, civil GPS signals are unencrypted and unauthenticated, which facilitates the execution of GPS spoofing attacks. During these attacks, adversaries mimic the authentic GPS signal and broadcast it to the targeted UAV in order to change its course, and force it to land or crash. In this study, we propose a GPS spoofing detection mecha…
Related Broader context
Fast Vehicle Detection in Aerial Imagery
Jennifer Carlet, Bernard Abayowa · 2017 · Computer Vision and Pattern Recognition · preprint, not peer reviewed · cited 19 times · AI summary available

The aerial-imagery study detects vehicles in images, whereas AircraftWitness localizes audio recordings by matching aircraft sounds with flight records.

Abstract
In recent years, several real-time or near real-time object detectors have been developed. However these object detectors are typically designed for first-person view images where the subject is large in the image and do not directly apply well to detecting vehicles in aerial imagery. Though some detectors have been developed for aerial imagery, these are either slow or do not handle multi-scale imagery very well. Here the popular YOLOv2 detector is modified to vastly improve it's performance on aerial data. The modified detector is compared to Faster RCNN on several aerial imagery datasets. The proposed detector gives near state of the art performance at more than 4x the speed.

Go further

How are these papers found?

We compare the title and abstract of this paper with those of every paper summarized on this site, and with the papers Semantic Scholar recommends, those it cites and those citing it (for a much-cited paper, its most recent citations and, from OpenAlex, the most cited ones). Among the prior work and the follow-ups, the papers that use each other in depth, a method it uses and the much-cited follow-ups come a little higher. The levels (very close, close, related) come from the similarity of the texts, not from a judgment of quality; * marks a paper without an abstract, compared by its title. The sentence under a paper is written by an AI from the two abstracts: check it in the papers. A paper marked not relevant by three readers leaves the list.

Updated on 9 November 2026.

Disclaimer: The AI-based summarization tool and virtual assistant provided on this website may not always provide accurate and complete summaries or responses. We encourage you to carefully review and evaluate the generated content to ensure its quality and relevance to your needs.

AircraftWitness: How Often Aircraft Can Locate Recordings

The essentials

Question

Method

Findings

Limitations

Why it matters

Read further

Definitions

Can an Aircraft Reveal Where a Recording Was Made?

A Test Planned Before the Audio Was Examined

Results Depend on the Evaluation Setting

GPS-timed recorder network

Randomly selected public recordings

What the Findings Do—and Do Not—Show

A Promising Clue, Not a Universal Locator

Ask questions about this paper to our AI assistant

Tools for this paper

Advanced tools Premium

Papers around this one

Closest papers

Go further

Cite this paper (BibTeX)