In early November 2026 the security researcher and YouTuber Colsto posted a video with an unsettling claim: the faint drone of planes in the background of a recording can reveal where it was made. He says he located a security-camera clip to within about 100 metres from its audio alone. No GPS, no landmarks, just a few passing planes. How can a plane’s hum do that?
Try it yourself in Flyover Doppler Lab: slide away from a plane’s flight path and watch the drop in its hum stretch out.
Why the hum drops
As a plane comes toward you, each sound wave it sends starts a little closer than the one before, so the waves arrive bunched up and the pitch sounds higher. As it flies away they arrive stretched out, and the pitch sounds lower. That is the Doppler effect. Take a turboprop on approach at 70 m/s whose propeller hums at 100 Hz: coming, you hear about 126 Hz; going, about 83 Hz. Those two ends depend only on the plane’s speed, not on where you stand.
The shape is the distance
What does depend on where you stand is how fast the pitch falls between them. Right under that plane, 300 metres up, half of the drop happens in about 5 seconds. Stand 1.5 km to the side and the same fall takes about 25 seconds: the plane is never closer than 1.5 km, so the angle it makes with you changes slowly. In the model the drop’s length is about 1.15 times your distance divided by the plane’s speed, and the halfway pitch arrives at the very moment the plane is closest. Fitting this curve to estimate an aircraft’s speed and altitude from the ground is an established acoustics technique; Colsto turned it around to find the listener.
Flight tracks are public
Aircraft with ADS-B broadcast their position, altitude and speed once a second, unencrypted, and networks of hobbyist receivers collect and archive them. If you know which plane you heard, its speed and its track, the drop tells you how far from that track you were: somewhere on a line either side of it. A second flyover from another direction gives a second pair of lines, and they cross at four spots; the time between the two drops is what picks one. Colsto says he matched two flyovers 920 seconds apart, and the silence of a third flight, against about 50 GB of US flight data, after months of work and more than 5,000 lines of code that he has not released. His 100-metre figure is his own claim and hasn’t been independently checked.
What the model leaves out
Flyover Doppler Lab assumes a straight, level flight at a steady speed through still air, one engine tone and a listener on flat ground. Real air has wind and temperature layers that bend sound, real planes curve, climb and slow down, and buildings echo. A real search fits the whole curve against many candidate flights and places, with terrain and weather. The tool explains the physics; it doesn’t analyse audio or locate anyone. Still, it is a useful reminder that background sound in a video you post can carry more information than you hear.
The film opens with a phone video of a jet passing low over a New York street (While on my Journey on YouTube; the pitch card over it is the model, not that plane’s recording), and uses two short excerpts of Colsto’s own video, a spectrogram of real flyovers and his map of one flight’s track, all credited on screen. The waves, pitch curves, map and numbers are drawn from the model.
Sources: Colsto, I discovered a sound that leaks your exact location (and it’s everywhere) (YouTube, November 2026); TechRadar, YouTuber discovers aircraft noise trick that can geolocate your videos to within 100m (3 November 2026); FAA, ADS-B: Ins and Outs; Wikipedia, Doppler effect and Automatic Dependent Surveillance–Broadcast; B. G. Ferguson, “A ground-based narrow-band passive acoustic technique for estimating the altitude and speed of a propeller-driven aircraft”, J. Acoust. Soc. Am. 92 (3), 1403 (1992).
Sources & further reading
Make the next question an experiment.
Browse independent creators who make big ideas visible.
Find an explainer

