Discussion
Holy shit.
This guy takes a couple of minutes of audio from a security camera anywhere in the US, with no clues from the video whatsoever, and finds the exact spot where it was recorded.
How?
By correlating the Doppler curve sounds from two almost inaudible planes passing by with public ADS-B flight data.
https://www.youtube.com/watch?v=UytYrhYLOKI (29 min)
I think this is the #OSINT video of the year.
YouTube
@scy so we juste have to stop every plane seems to be a good idea
By the way, on the same channel there's also this video: Finding a location by utilizing thunder sounds in the audio's background
https://youtu.be/SnbkIm56TfI (15 min)
Because it turns out that we as a civilization record every single lightning strike's time and location.
https://en.wikipedia.org/wiki/Lightning#Detection_and_monitoring
YouTube
In the UK the police record the hum from the electrical grid and have used it in court to prosecute criminals.
https://www.crimeandinvestigation.co.uk/articles/hidden-electrical-noise-can-catch-criminals
@scy american woukd use anything but the gps system
(impressive work)
@scy the maths behind this is so cool, it's a really good video
@scy Interestingly this wouldn't work in a lot of Ukraine. No flights. Well, not commercial traffic.
@scy but, if you have the audio from two cameras, which are presumably fixed in position, then you have already narrowed the area down quite a lot.
@scy Excellent method if you happen to have misplaced your security camera.
@scy the problem with security on the internet is, you have to consider future attacks on what you uploaded, that don't exist yet.
@scy ah, thats how, thanks! I have this in a private tab but I've not watched it yet.
@scy This same guy also did this with thunder.
@scy I wonder if me living near an international airport with near constant air travel means this is easier to do or harder to do to my audio recordings
I would assume that any municipal microphone (ALPER, ShotSpotter, security camera) could map a city and provide ground truth for anonymous audio.
@scy @drahardja I guess you have to do your shenanigans near airports - or fake an airport by playing random doppler curves… Impressive!
@scy Military forces have been doing this for decades.
@scy Man..., and my high was just hitting that good place too...
@scy That is incredible.
The physics modelling, searching over massive data sets, layering all of that together to find a needle in that haystack.
@scy Amazing. What a thrilling video. Thanks for sharing.
@scy sometimes I think about all the things we’re going to learn from old film (eg, Zapruder) with mass facial recognition and things like this. It’s an amazing time.
Two thoughts about this method:
1. They assume a constant base frequency that only shifts because of the Doppler effect, but I'd argue you can get more precise results if you consider frequency changes due to flight maneuvers, changing speed etc., too.
2. You could combine this method with ENF analysis to narrow down the date and/or time.
https://en.wikipedia.org/wiki/Electrical_network_frequency_analysis
YouTube
@scy Well, he does compute the frequency changes for each pair of flight and candidate point, and the flight data includes the flight path's geometry and changes in speed. Isn't that what you mean? 🤔
@scy I’m surprised he didn’t have to include wind in the model. I wonder if this is just luck or systematic.