Aircraft Doppler geolocation projects after colsto's video
46 open projects and one forum tool that try to locate a recording from propeller-aircraft Doppler curves matched to ADS-B tracks. Checked 5 November 2026.
The method comes from colsto's video I discovered a sound that leaks your exact location. Of these 47 entries, 21 report a correct location on a real recording. 10 did it only on the video's own recording, 7 only on other recordings, and 4 on both. The other 26 report no real-world success.
11 projects also listen for sounds other than aircraft. Thunder and mains hum are the most common. Thunder, church bells, blasts and emergency-alert bursts can point to a place. Mains hum mostly narrows the date, time or power grid.
| Name | Created (UTC) | Real-world success | Other sound sources | Additions or changes from the video |
|---|
Sounds other than aircraft
These projects use the following non-aircraft sounds:
- thunder timed against lightning-strike records: Earwitness, Audio Fly-over Geolocation, GeoSonnar and TrackSound. No project reports locating a real recording this way
- mains hum (electric network frequency) for the power grid, date or time: Provenance, TrackSound, jmarcoz1, flyover-locator, Mearman/doppler and colsto-leak-and-mask. Provenance also reads it from light flicker in video
- church bells compared with location-tagged bell recordings: EchoTrace
- wind turbines, screened only: EchoTrace
- quarry blasts, explosions and sonic booms, and EAS or NOAA Weather Radio data bursts: TrackSound
- siren style as a country hint: colsto-leak-and-mask
- Doppler from cars, trucks and boats, labelled but not matched: flyover-locator
- general sound labels from the YAMNet classifier: Soundleak
Helicopters count as aircraft here. Several projects handle them, including doppler-geo, Overhead and TrackSound.
How the success column was judged
- video means the project reports a correct location on colsto's own recording (Oyster Bay, New York, 8 September 2026)
- other means it reports a correct location on a different real recording
- the bar is the project's own claim. We did not run any code
- a result the author rejected counts as no. This applies to DOPPLERGEIST (44 m, rejected) and the SokolskyNikita fork (rated ambiguous by its own tool)
- the notes give search area, time knowledge and tuning, because successes differ a lot in how much help they had
Weakest successes
These would drop to no under a stricter bar:
- colsto-leak-and-mask, Soundleak and Legendile7 reused another tool's measurements or the true flight tracks
- Doppler Dox and OVERPASS searched boxes chosen around the known answer
Sources and coverage
- dates are GitHub repository creation times in UTC. Sitrec's date comes from the Metabunk thread, which shows only relative times
- we found the projects through GitHub repository and code search. Projects without "doppler" or a link to the video in their text may be missing
- left out: a copy of digitalis-p/doppler-locate, a detector-only repository, a README-only repository, an old snapshot, 2 unrelated repositories, 3 empty ones and 8 tools that only remove aircraft tones