Slashdot Mirror
Electrical Grid Hum Used To Time Locate Any Digital Recording
illtud writes "It appears that the Metropolitan Police in London have been recording the frequency of the mains supply for the past 7 years. With this, they claim to be able to pick up the hum from any digital recording and tell when the recording was made. From the article: 'Comparing the unique pattern of the frequencies on an audio recording with a database that has been logging these changes for 24 hours a day, 365 days a year provides a digital watermark: a date and time stamp on the recording.'"
Deja vu:
http://science.slashdot.org/story/12/12/12/1331243/engineers-use-electrical-hum-to-fight-crime
...application of technology and not one utterance of "boffin"??!?!?
I demand satisfaction!
http://science.slashdot.org/story/12/12/12/1331243/engineers-use-electrical-hum-to-fight-crime
systemd is Roko's Basilisk.
Re: Location: Wrong. The entire UK grid is "locked together" and it all runs at the same frequency. Necessarily. Also: Recorder doesn't need to be plugged into the mains. 50Hz hum permeates the space around us. Try grabbing hold of an oscilloscope lead and look at how much 50Hz hum you are "carrying". Unless you're a long way from mains outlets, it's a lot.
"Absorbing your worst..."
At the utterly fascinating Georgetown Steam Plant Museum in Seattle, I learned of the difficulties in getting a (somewhat elderly) generator in sync with the grid. Apparently, get it right and all the other power stations will pull it into the exact frequency - get it wrong, and you'd snap the turbine shaft.
As for the mains hum, in an undergraduate experiment at Jodrell Bank Radio Observatory, I detected intelligent life - on Earth, unfortunately. While running an FFT on a recording of a pulsar, we not only uncovered the spinning neutron star's rotation - we also discovered some not-exactly-mysterious peaks at multiples of 50Hz.
Tedious Bloggy Stuff - hooray?
You can hear 50hz. It won't be stripped out, although it's such a low volume noise you'll need some funky recovery algorithms to pick it out - oh look, those have been around for a while now. I'd be interested to hear exactly how accurately they can pinpoint the time and date. Are we talking to within a few hours, or are there sufficiently frequent irregular fluctuations you can do pattern matching to pin it down precisely to the second?
I do have my doubts over how resiliant this technique would be to forgery. If the police can record the hum, so can human beings. Say you wanted to have a conversation with someone verified by police as taking place after it really did. You record the conversation, use live hum data to cancel it out without damaging the audio, then a week later you record and mix in some fresh live hum noise. Can't see any decent sound engineer with the right equipment having any trouble with that, I know a guy who'd have a whale of time with it, and there goes any hope of this evidence ever standing up in court.
Nah, Read The Fucker...
Tomorrow is another day...
You're not giving them due credit for their pioneering use of amazing new load-balanced submissions technology, in which each word of a post is sent to a different editor for review.
Sure, they still have a few niggles to sort out, but man, this is the future!
Now here's an idea for an even cooler application: A web service which allows customers to upload any edited audio recording and I can apply a subtle hum with a user-selected timestamp so it authenticates as "not edited original recording" with the Met Police's database! I shall start recording the mains hum shortly. Criminals rejoice! Huahahahahaha!
well. some people can hear it.
back when I was in uni the multimedia lecturer was playing tones at different frequencies. "oh, and any of you who've spent too long in the computer lab won't be able to hear this" most of us were stone deaf in a small range around the frequencies put out by electrical equipment.
Release the Ferrets!
This space available.
Well, I'm sure that the 50Hz hum sounds much warmer in its original format without digital sampling errors and whaarrrgarrbbll bitrate rhubarb rhubarb harmonics frequency blah-diddy blah blah blah oxygen-enhanced one-way digital cables, so there!
They sentenced me to twenty years of boredom
As with so many of these forensic tests, what this test can unambiguously be used for is to prove that a recording _HAS_ been tampered with.
If there is no hum, or it appears to be correct then the most you can say is "it might be a genuine recording made at the appropriate time" but if there is hum, and it's obviously discontinuous then you can categorically state that the recording has been tampered with.
Unfortunately, most people, including prosecutors, defence, juries, judges, politicians etc, do not understand the distinction. There are two possible answers to "has this recording been tampered with:" YES and I DON'T KNOW. People like certainty so this gets changed to YES and NO and while in most cases that NO does turn out to be correct, you get miscarriages of justice when it's not the case.
Tim.
God said, "div D = rho, div B = 0, curl E = -@B/@t, curl H = J + @D/@t," and there was light.
You can see the current state of the UK power generation network dynamically here:
http://www.dynamicdemand.co.uk/grid.htm
I use 60Hz you insensitive clod!
In that case you probably are in North America.
Now they know where you are and when. Other information will be available shortly and you won't be anonymous for much longer...
I'll see your Constitution and raise you a Queen.
I do have my doubts over how resiliant this technique would be to forgery. If the police can record the hum, so can human beings. Say you wanted to have a conversation with someone verified by police as taking place after it really did. You record the conversation, use live hum data to cancel it out without damaging the audio, then a week later you record and mix in some fresh live hum noise. Can't see any decent sound engineer with the right equipment having any trouble with that, I know a guy who'd have a whale of time with it, and there goes any hope of this evidence ever standing up in court.
I am not familiar with removing noise in the audio spectrum but am in the RF spectrum. If we have a predictable noise source it can be filtered out, the problem is that when you take out the hum you will also take out some of the other noise as well, then when you put the correct hum in it's not interacting with the same noise frequencies, so you are left with nulls in the frequencies you removed the hum from which may be detectable. This eliminates putting the correct hum onto a recording made at a different time. The other method is to dub a voice onto the recording, when you record the person you will also get noise in that recording, noise that has to be eliminated as it will be out of place in the final recording. Again when you remove the noise you will be removing part of the signal that you want to keep which may be detectable. The only real way to isolate the voice from everything else is to record in a quiet room as trying to isolate the signal from the noise becomes difficult, so you would need to get the victim to cooperate. While I agree that it's not impossible to forge a recording, if the listening device is good the noise becomes too hard to forge.
Knowledge = Power
P= W/t
t=Money
Money = Work/Knowledge so the less you know the more you make
Okay, fine, so it has a 50hz beat. But but the question is, can you dance to it?
Slightly off-topic but fun: my favorite party trick for a self-confessed audiophile -- Admire their overpriced collection of obscure amps and equipment costing more than the average suburban home. Put on a look of concern and when asked what's the matter, mutter "Lovely bit of kit, andand perhaps it's just my ears, but I think there's some muddiness in the midrange that erodes the transparency." Off they go, down the rabbit hole of knob twiddling and speaker placement for the next several hours.