all 2 comments

sorted by: hot top controversial new old
[–] 4 points 4 days ago

This video is awesome. Thanks!

  • source
  • [–] 1 point 4 days ago

    I can't believe how smart some people are

  • source
  • 7 comments in c/privacy@lemmy.ml view that post
    [–] 20 points 2 days ago* (1 child)

    Haven't seen this video but I think Veritasium or Tom Scott did a video about this like a decade ago. They got creepily accurate results. Mains hum be scary

  • source
  • hideshow 1 child comment
  • [–] 23 points 2 days ago

    This particular method is triangulating background aircraft doppler sounds, getting their approximate speed and distance from the doppler shape and looking it up in air traffic logs to find valid matches. it needs multiple data points though.

    it's not perfect of course, there are plenty of light aircraft that dont have accurate tracks available, and it would be easy to spoof or mask, but it's very cool bit of work that could be added to the toolkit for people that really want to use it.

  • source
  • parent
  • [–] 1 point 2 days ago* (last edited 1 day ago) (4 children)

    Could someone just play a pre-recorded loop of those kinds of subharmonic sounds at low volume in the background of their video in order to mask the real Doppler curves and such in the audio?

  • source
  • hideshow 4 child comments
  • [–] 8 points 2 days ago (3 children)

    I mean yeah. But you can also just filter out the frequencies in post.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 10 points 2 days ago (1 child)

    But the filtered out low frequencies will have modulated the higher frequencies of the the intended audio content. More difficult to process but the LF signal is there.

    Layering false signals on top just means there's fake locations to rule out.

    And if it's video: https://news.mit.edu/2014/algorithm-recovers-speech-from-vibrations-0804

  • source
  • parent
  • hideshow 1 child comment
  • [–] 3 points 1 day ago

    They go through this in the video. The frequencies in question overlap with large portions of the vocal range, so not very practical. They also show that it’s pretty easy to recreate the data even when it’s attenuated.

  • source
  • parent
  • 7 comments in c/opsec@lemmy.zip view that post
    [–] 13 points 2 days ago* (last edited 2 days ago) (5 children)

    Watched this video a few days ago, so insane. Basically every live streamer will be doxed at this point

  • source
  • hideshow 5 child comments
  • [–] [S] 7 points 2 days ago (4 children)

    And this is probably a crude method compared to the ones we don't know about yet, but it was still accurate to some ~30 feet I think he said? Insane! This guy also located places by thunderstorm videos so if there's no planes around but its raining that's enough to get locations

  • source
  • parent
  • hideshow 4 child comments
  • [–] 6 points 2 days ago (2 children)

    His whole channel is great paranoia fuel

  • source
  • parent
  • hideshow 2 child comments
  • Is this how we'll tell if a vid is "AI"-gen in the near future?

  • source
  • 1 comment in c/privacy@programming.dev view that post
    [–] 6 points 4 days ago

    Thanks for building this torment nexus, guy

  • source
  • 1 comment in c/privacy@lemmy.ca view that post
    [–] 4 points 1 day ago

    That's actually really cool! Glad they went through the math of it.

    But that said, it would be a lot harder to map using this method accurately near airports and military bases since some planes do not activate ADS-B. The method also requires at least the exact date of filming, even if you could get flight data from across the world.

    TL:DW; say "eeeeeeeeowwwwwwww" at random intervals on videos to mask your location being identified by background plane sounds /s

  • source