QotNews Hacker News, Reddit, Lobsters, and Tildes articles rendered in reader mode.
YouTuber discovers inaudible aircraft noise can geolocate your videos to within 100m — 'Doppler effect' pinpoints locations from just 60 seconds of audio
I remember an old whodunit finding a general location by listening to trains on a recording and knowing the timetable. This seems like a more sophisticated version of that.
The story of Shia LeBeouf's He Will Not Divide Us installation has some pretty incredible stuff like this. He moved the work to secret locations, and people found it by things like correlating flashing red lights on the wall with known ambulance callouts.
This new thing seems worse though, because there's no human effort involved, just feed the audio to some compute.
Most of that was bullshit though. People had a general idea of where to look immediately because of where he had been seen.
Afterwards all these crazy stories popped up that made it all sound a whole lot more Sherlock Holmes than it actually was. But it was mostly people pulling shit out of their ass about a location they already heavily suspected.
An enormous amount of information about airplane flights are needed for this technique to work, I wouldn't be surprised if there's a relatively recent cutoff that this isn't usable before.
Sure. I dunno if you watched the whole video but the analysis appears robust enough to be able to identify a Doppler curve that doesn't match any public data and then exclude that from the scoring. It's more that there's generally enough air traffic that is recorded on any given day. The whole thing is sort of dependent on knowing roughly when a video was recorded though, or at least the amount of processing required otherwise becomes enormous.
Sure, computationally it's not that complex, I'm just saying the search space can grow very large depending on how much info you have about the time or location, or how many doppler curves the recording captures.
So I'm not sure if this false description of the Doppler effect is in the video or only TechRadar got it wrong, but it's annoying and makes me doubt the claims:
> As Colsto explains, the key is the 'Doppler effect'. That's the way sounds appear to change based on the distance between a moving source and the listener.
No, that's not the Doppler effect. It has nothing to do with distance, only with relative velocity. As the ambulance goes by, its relative velocity component towards you changes, which changes the pitch.
No I didn't watch it yet, it just immediately ticked me off but I'll watch it later when I have a few minutes. I don't need the principles explained, I understand how this could work in principle, but if the basics are off I get my doubts about whether it's real or a publicity stunt or whatever.
Credible for a stationary outdoor recording with identifiable aircraft tones, usable historical tracks and timing constraints. Much less dependable for an arbitrary short indoor or noisy clip of unknown date. The privacy implication is real, but 'any video can reveal your exact location' goes well beyond what this demonstration establishes.
Ancient_Skirt_8828 | 13 hours ago
I remember an old whodunit finding a general location by listening to trains on a recording and knowing the timetable. This seems like a more sophisticated version of that.
Space_Pirate_R | 10 hours ago
The story of Shia LeBeouf's He Will Not Divide Us installation has some pretty incredible stuff like this. He moved the work to secret locations, and people found it by things like correlating flashing red lights on the wall with known ambulance callouts.
This new thing seems worse though, because there's no human effort involved, just feed the audio to some compute.
ClarifiedInsanity | 8 hours ago
Most of that was bullshit though. People had a general idea of where to look immediately because of where he had been seen.
Afterwards all these crazy stories popped up that made it all sound a whole lot more Sherlock Holmes than it actually was. But it was mostly people pulling shit out of their ass about a location they already heavily suspected.
OurDumbCentury | 9 hours ago
Someone also drove around honking their car horn to audio locate as well.
Renarudo | 6 hours ago
Perfect, since my trains are frequently delayed, they’ll never figure it out
Ancient_Skirt_8828 | 5 hours ago
Are you planning to kidnap someone for ransom?
Phine420 | 2 hours ago
No, for fun
Giffdev | 10 hours ago
Isn't that in the fugitive where they figure out it's the elevated train in Chicago
axl3ros3 | 5 hours ago
Made me think of Blow Out (1981)
Not the exact same but adjacent
userjc247746 | 14 hours ago
✅ Makes all videos 59 seconds or less.
Kahnza | 13 hours ago
Bring back Vine! Ultra secure!
OminOus_PancakeS | 10 hours ago
🎶 Ah love youuu... biiiiitch
americablanco | 10 hours ago
I ain’t ever gonna leave ya… biiiiitch
ggrieves | 12 hours ago
Reminds me of that movie trope where the police had to keep the kidnapper on the phone long enough for a trace to complete
Vishnej | 11 hours ago
Colsto, this Youtuber, has already been doing pretty amazing unrelated things with geospatial OSINT.
Charlos11 | 13 hours ago
Thanks I hate it.
FaceDeer | 11 hours ago
Would you rather not know this was possible? I expect most of the three-letter agencies were already aware of this technique.
Buffalo-2023 | 13 hours ago
maybe use this to find cold kidnap cases where there is a video demand
FaceDeer | 11 hours ago
An enormous amount of information about airplane flights are needed for this technique to work, I wouldn't be surprised if there's a relatively recent cutoff that this isn't usable before.
Photodan24 | 13 hours ago
It assumes that every prop plane that's audible is equipped with ADS-B, which is not true by any means.
fatcatfan | 13 hours ago
Sure. I dunno if you watched the whole video but the analysis appears robust enough to be able to identify a Doppler curve that doesn't match any public data and then exclude that from the scoring. It's more that there's generally enough air traffic that is recorded on any given day. The whole thing is sort of dependent on knowing roughly when a video was recorded though, or at least the amount of processing required otherwise becomes enormous.
ImpetuousWombat | 11 hours ago
Computers excel at this kind of processing, and could search for candidate time periods from a flight database on low-end consumer hardware just fine
fatcatfan | 11 hours ago
Sure, computationally it's not that complex, I'm just saying the search space can grow very large depending on how much info you have about the time or location, or how many doppler curves the recording captures.
bstabens | 6 hours ago
I don't know. The more doppler curve, the closer to an airport you are I'd guess. Which makes it more distinct flight patterns, so to say.
While somewhere far out, you'd have to compare *more* timetables since you don't know where the planes you hear are from.
Illlogik1 | 13 hours ago
Wait till skynet gets its hands on this !
arealuser100notfake | 13 hours ago
The moment they published the article it did
motorhead84 | 11 hours ago
Skynet probably published the article just to rub it's superiority in our faces.
MezoDog | 13 hours ago
Science!
PhD_Pwnology | 10 hours ago
Jokes on them, I edit my videos so all backgrounds noise layers are fat bastard from Austin Powers farting.
darkness_calming | 7 hours ago
Several lines of code?
reallifedog | 12 hours ago
Isn't that how Mark Hoppus (from Blink 182) helped find Osama Bin Laden?
PinothyJ | 9 hours ago
It says the audio is inaudible. So, does that means it is low frequencies? Would a high pass on 30Hz solve it then?
kallicks | 12 hours ago
Goodluck around Dreamland guys.
Elehir | 11 hours ago
2b2t thunder exploit IRL
nationunderfraud1 | 11 hours ago
So there has to be an airplane nearby or captured on recording
This-Construction926 | 9 hours ago
If this is one guy, imagine with the NSA/CIA could do
2Throwscrewsatit | 6 hours ago
Good thing I have a fan on next to me whenever I record. Good luck!
Langdon_St_Ives | 13 hours ago
So I'm not sure if this false description of the Doppler effect is in the video or only TechRadar got it wrong, but it's annoying and makes me doubt the claims:
> As Colsto explains, the key is the 'Doppler effect'. That's the way sounds appear to change based on the distance between a moving source and the listener.
No, that's not the Doppler effect. It has nothing to do with distance, only with relative velocity. As the ambulance goes by, its relative velocity component towards you changes, which changes the pitch.
rustajb | 13 hours ago
Did you watch the video? this is explained in detail and it's not difficult to follow.
Langdon_St_Ives | 13 hours ago
No I didn't watch it yet, it just immediately ticked me off but I'll watch it later when I have a few minutes. I don't need the principles explained, I understand how this could work in principle, but if the basics are off I get my doubts about whether it's real or a publicity stunt or whatever.
mycall | 9 hours ago
Credible for a stationary outdoor recording with identifiable aircraft tones, usable historical tracks and timing constraints. Much less dependable for an arbitrary short indoor or noisy clip of unknown date. The privacy implication is real, but 'any video can reveal your exact location' goes well beyond what this demonstration establishes.
hellsbellsvr | 12 hours ago
Jokes on him —- my house without any airline traffic overhead.
FaceDeer | 11 hours ago
Joke's on you, it doesn't have to be overhead. Maybe watch the video.