An inaudible aircraft trace located a video within a few hundred feet

Colsto started with an almost static video, the date September 8 and the knowledge that it was recorded in the continental United States. He extracted two faint Doppler curves from propeller aircraft in the spectrogram and matched them with public ADS-B records.

Two aircraft 920 seconds apart narrowed a search across the United States

The first pass found places where an observer could have heard one suitable aircraft and another 920 seconds later. The calculation accounted for speed, altitude, terrain and the time sound needed to travel. Silence between the two events became another filter.

Colsto then modeled the precise Doppler curve produced by real flight paths. The best match led to a riverbank where Street View confirmed both the tree line and the camera on a pole. He describes the resulting precision as a few hundred feet.

Background audio becomes another layer of location metadata

The main demonstration used a dated video and flight records. A separate test located the same place from one minute of a nearly inaudible aircraft trace when the exact recording time was known. That matters for calls, security footage and videos without visual landmarks.

For OSINT, aircraft audio joins shadows, weather and traffic as a practical clue. For journalists, investigators and vulnerable sources, it is a warning that stripping GPS data does not remove every location signal from a recording.

The attack still needs data, computation and unusually suitable audio

Colsto describes months of work, more than 5,000 lines of code and 50 GB of data. Missing timestamps, incomplete flight records or a weak trace make the search far harder. He also notes that an attacker will usually have easier ways to locate someone.

A simple high-pass filter may fail because the relevant frequencies overlap human speech, and the trace can survive phone compression. This is a capability demonstration, not an automatic tool that reliably locates every recording.

Practical tools will show whether the technique leaves one laboratory

The important next step is a repeatable tool that can ingest audio and available ADS-B records without months of manual tuning. Failure cases need equal attention, including wrong timestamps, missing flights, urban noise and different microphones.

There is no simple defense switch yet. Edits, targeted denoising and withholding precise timing can make a search harder, but any treatment must preserve intelligible speech and the evidentiary value of the recording.

Lilith's verdict

The camera stood by the water and showed almost nothing. Two aircraft that never entered the frame still gave it away. Privacy can leave through the microphone while everyone watches GPS.