Police scientists have hailed a new technique that recently played a pivotal role in securing a murder conviction as the most significant development in audio forensics since Watergate. The capability, called "electrical network frequency analysis" (ENF), is now attracting interest from the FBI and is considered the exciting new …
All sounds a bit analogue to me
Hmm, does this only work on magnetic media? it does not say so in the article, but I have a feeling that it does, if so it may be of limited lifetime?
I suppose there is digital static, but id imagine its more likley to be muted out by a recoding device, especially anything using compression?
Re: All sounds a bit analogue to me
"Hmm, does this only work on magnetic media?"
No, it only works on digital media, as the article says:
"ENF has basically been made possible by the move to digital recording," Dr Cooper said.
"Old magnetic cassette and VHS tapes didn't keep time accurately enough to extract reliable data, but now we can analyse even cheap voice recorders."
Must pay more attention
If I happen to have my home equipment running off a UPS which is filtering the power then this technique is useless because nothing is subjected to the tender mercies of the local power grid?
As the article says, this will even work on battery powered devices as long as they are in the vicinity of mains power.
I suppose it may be possible to climb into a Faraday cage or go somewhere where there is no mains electrickery nearby.
I wonder if using a generator or inverter to power the recording device, or generate a powerful enough field obscure that of the nation grid, would be sufficient to defeat this technique.
I suppose that an enterprising crook could record the variations in mains freq themselves and then apply them to a recording though It would be difficult to make this appear authentic, particularly if the analysis takes harmonics into account.
Or somewhere remote like the Tora Bora mountains.
call me paranoid
But surely it would be trivially easy to strip those ENF frequencies from a piece of audio, or even to mix a recording of ENF made at a different time in to an existing recording?
Yes, you're paranoid...
But imagine a terorrist, lets call him Osama Been Late or something, sends a 'new' video of himself to a local newsstation, threatening to pollute the World's stores of Jarlsberg cheese with cheag Edam from the USA?
With the ENF tampered with the Keystone Cops can say that 'hey, this is an edited video, the guy behind the threats haven't made a reliable appearance in yoinks, and is probably dead'
In fact, it's in the terrorists own interest to keep the ENF as some sort of 'Open Source Digital Fingerprint' to prove that they are alive, well, and hoping that everyone else isn't.
(And if the ENF signature matched Acapulco, Las Vegas or another nice vacation location... they're obviously very well... )
OK - You're paranoid
Well yes - but wouldn't that be self-defeating in this case? In the article, the entire point of correlating the ENF on the tape with the database was to prove the authenticity of the tape and to demonstrate that it had NOT been edited. If you strip out all the ENF, then you have, by definition, altered the original, which may invite questions as to its authenticity.
Remember this next time you see the news.
I think you might be a bit oversensitive
There's really no need to comment on every article that mentions the word "Romania" unless you have something useful to say.
So, basically, it's a Spanish invention, right? Old Spain created the modern Romania.
Er, actually I think you'll find it was the Romans (the clue's in the name) who, last time I looked, came from what is now Italy rather than Spain.
The old kingdom of Dacia (as it was called) had been something of a thorn in Roman sides for many years, conducting an on/off war with Rome. A combination of the terrain and their troops (drilled heavy armoured foot with two-handed heavy weapons) was rather unfavourable to the Roman Legion system. When the Emperor Trajan finally stuffed 'em, he renamed the place "Romania" to send a not-so-subtle message to any other places that might be entertaining ideas of sticking it to the Empire.
"renamed the place" rather understates the facts here. He either carted off to slavery or exterminated every human being he found living in the place. After he'd finished, the only human residents were Roman colonists. I suspect the name change was the *most* subtle of the messages that he sent to any other places.
Sorry, you are right. I get carried away sometimes :)
How do the fluctuations in mains power get transferred to the recordings? Mains-powered devices go to significant effort to remove what used to be called "mains hum" and it doesn't seem likely that battery-powered devices would be very susceptible to local mains interference to the extent that "ENF" data would actually get into the recordings. Presumably the ENF information if present would appear in audio recordings as an approximately 50Hz pattern and I don't recall ever seeing anything like that.
It is the pattern that is important. How variations in mains frequency affect a recording device are probably more complex than merely inducing 50Hz mains hum onto the recording.
Variation in mains frequency may also induce harmonics, vary sampling speed and/or quantisation levels. As long as the variation in mains freq for a given time frame = the variation in the measured attribute then you have a match.
Well it obviously only works when you have a little bit of line hum in your recordings. If you filter that out, there's nothing.
Besides you need to keep in mind that they probably use corellation, so they can easily boost the signal to noise ratio by a factor of 1000.
They go to great pains to dampen it enough that it's inaudible.
But just think how many devices around you TRANSMIT that noise?
And not always as audible or electromagnetic noise?
Ever thought that maybe that lightbulb in the corner flickered a little?
Get a good enough recording of it and that too can be read out and matched with a signature.
And THAT will also go through a Faraday cage...
light bulb flicker?
I doubt very much that consumer grade digital recording equipment would be sensitive enough to pick up the noise of a light bulb flickering... but amp it up and look for specific harmonics of the 50hz signal, maybe.
sounds quite straightforward
electricity, AC in particular, induces electromagnetic fields that propagate outwards
microphones, convert movement into electromagnetic fields that are recorded. It's quite plausible to believe that modern digital recordings will record minor fluctuations in nearby electromagnetic fields as "noise".
I'd expect any sort of lossy compression to filter it out though.
What a load of old...
Sorry, but that just sounds like something off The Avengers or Sapphire and Steel.
I call BULLSHIT on that one. Utter bullshit. Never heard such a load of old cock wipe. Utter fantasy.
I suppose you don't believe they can bounce lasers off of glass windows to capture the sound waves that are passing through the glass and thus disrupting the light waves as they diffract either.
Or that they can send broadband and hi-fi audio down a live mains cable? Or that ADSL can work over a normal phone line, which as any fule kno, only goes up to 64kbps. Who do BT think they're fooling when I download a file at 8mbps? Not me, I can assure you! I don't care that a 6mb file appears on my desktop in one second, common sense tells me I'm only downloading at 64kbps because I canne change the laws of physics, Captain!
Now if you'll excuse me, I'm just off to arrest a black man for urinating in a public convenience and looking at me in a funny way.
Mine's the one in the tin foil hat.
Re Flat Earthers!
"bounce lasers off of glass windows to capture the sound waves..." I'm still waiting to see this in practice (i.e. the real, rough world out there) and not just in a lab or on TV.
Re: Re: Flat Earthers
Since laser microphones are usually used for covert surveillance you are unlikely to see them in use. You could always search the Internet for "laser microphone", buy one and then use it.
or build one yourself
a laser microphone that is:
Re: "and not just in a lab or on TV"
"At the Metropolitan Police's digital forensics lab in Penge...."
So this one's just in the lab then and hasn't even made it to TV yet. Looks like laser mics are ahead in this game.....
A new spin-off
It lacks a certain glamour, though.
Actually you can play back VHS very accurately. The whole timing is controlled by the video signal which is usually derived from a quartz. Now if you play it back at the right speed, you will have very precise timings. What you need for that is a propper VTR you can sync to external syncs. Or you can use the obvious cheaper ways.
It does sound like BS
A digital recording done on consumer equipment - that isn't going to be that good anyway - isn't going to be able to accurately reproduce random background intereference with any accuracy.
I'd believe it more of analogue equipment but even then.. it wouldn't be enough you could detect anyway compared to the random RF crap that hangs around every house.
ENF database a secret?
if someone had the data they could tamper with a tampered audio clip to match a specific point in time!
Also, I could modify a recorder and wobble the quartz a bit with a 50Hz signal not derived from the mains, and add "plausible deniability".
...police declined to name the murderer, citing undisclosed "operational reasons"
Someone's using the file to prop up a table leg.
Absolutely brilliant idea
Surprising there are so many naysayers - this sounds like an idea that would work.
There is mains background hum on *any* audio recording. Yes, it's low level, but it's there.
And a record of the National Grid's random variation (50Hz +/- 0.5Hz, typically) provides an excellent timestamp.
I would doubt that they would be able to accurately date any random sample of audio, But to prove that the audio sample was taken at a specified date and time? Sounds very reasonable to me.
A bit DARPA to me.
Is that a black twin rotor Osprey I see in the distance.
I must've missed something somewhere...
"Over a short time they form a *unique* signature of the electrical frequency at that time, which research has shown is *the same in London as it is in Glasgow*."
How does 'unique' = 'same in London as it is in Glasgow'?
Also, how good would this work if said ne'er do well simply stood in the middle of a rather large, very empty, very much disconnected, disused warehouse to make those all important phone calls. I can imagine these often make suitable venues for such nefarious activities.
Re: I must've missed something somewhere...
"How does 'unique' = 'same in London as it is in Glasgow'?"
Because it is unique in time, not space.
It's very important that the whole of the national grid is working on the same frequency, really bad shit (tm) happens if it's at different frequencies.
still don't get it
So if I record a short wav file using sndrec32.exe I should be able to open it in an audio editor and see some ENF? But I can't, so where is it?
They know what they are looking for, this isn't a 50Hz wave we are talking about, its disruptions to it.
You equipment is not sensative enough, and I asume that the noise is not background noise, but input changes. You probably have to do some quight complex filtering to get the infomation out.
nope nothing nada
You do not need sensitive equipment. Any and every audio recording device receives the ENF infiltration whether it wants it or not (according to the met.) But strangely the wav file I recorded on my no-microphone bog-standard PC is completely empty.
Why not read the original research papers?
Some of Dr. Grigoras' papers need to be paid for, but the pdf link below is free, and describes the method as it was around 2003. There are several papers on ENF by other authors in the Audio Engineering Society library, but again you'll have to pay for them.
Sensitive enough to view
No microphone ?
Then how can you be sure that the program is not just arbitrarily putting nothing in the recording since it doesn't detect any input ?
I don't mean to say there isn't any, I mean to say that the engineers that devised the program probably included some threshold barrier under which the decision is to simply record "nothing".
So you don't see any ENF variation because the program decided to write "nothing" instead of writing the actual nothing it was getting.
Get it ?
Or do I not get it ?
I'm confused now.
Won't name the case?
I find it strange that the Met refuse to disclose the details of the 'high profile' case unless there is some FUD going on here. If this was a new technique that helped secure the conviction surely it would be a matter of public record as part of the court case? Or did the police not disclose all their evidence at the time?
Whilst I can see that the basic theory of the technique is at least feasible, I'm a little more skeptical about the claim that the recorded pattern would be the same from London to Glasgow. Surely local substations and transformers have some effect?
Re: Won't name the case?
"Whilst I can see that the basic theory of the technique is at least feasible, I'm a little more skeptical about the claim that the recorded pattern would be the same from London to Glasgow. Surely local substations and transformers have some effect?"
The research is here. Unfortunately only the abstract is free: http://www.aes.org/e-lib/browse.cfm?elib=14411
Same in London, Glasgow and a faker's audio lab ?
This is because the entire UK grid is synchronised. You wouldn't get the same variation pattern in London as in Paris because mains electricity is exchanged between the UK and France using DC and not AC. But you'd get the same pattern in Paris as in Frankfurt. And if the historical frequency variation data isn't already published somewhere on the Net it probably soon will be.
Let's hypothesise you wanted to create a faked signature and you had access to the historical data of the AC variation within the AC control region against base 50Hz frequency from the date and time you wished to simulate. You then transform the mains using a DC converter (e.g. go through a bank of 12V car batteries) and then modulate the local AC mains as output from the AC->DC->AC converter to the same time signature you wished to recreate. If we don't yet know whether this is possible, it would take further research to prove or disprove whether this approach could fool this forensic method. If it could, then someone wanting to claim a recording was faked would presumably have to demonstrate a probability that this approach was used to reinforce the faking of a date/timestamp on a recording.
While we will welcome use of such approaches to convict the guilty, this could be of interest to someone genuinely believing they were framed by this approach. Let's face it, history isn't short of examples of supposedly unchallengeable forensic methods found to have been used to wrongly convict the innocent.
If they were asked nicely enough
Would the electricity industry introduce some "signature" noise into their supply please so there is an unmistakeable data stream in the mains power?
Could they please include some data about real-time supply against demand, then we just need a gadget to read the data and use it for controlling various domestic appliances like freezers, water heaters and 'leccy car chargers, and so balance some of the load upon the grid.
Might be more acceptable to more people than rolling blackouts in a few years' time, given that the alternative is ferocious rises in the price of peak-time electricity.
EDF does introduce data
The lights "wobble". Especially fluourescent. But the people at work think I'm mad, they apparently cannot see it. But I can tell the time with it. 8pm, 10.30pm. Midnight... And other times through the day, the lights flicker quite alarmingly (for me, but then I find many older cinemas gave me headaches). I *think* this is data for switching equipment, like, night tariff etc. I can imagine that would be able to be picked up in a recording, and if the data being transmitted actually carries a datastamp...
BTW, if anybody has any information on what this 'flickering' is for (English or French), please reply with a URL. I'd like to know for show my cow-orkers that I'm not crazy. If I was, I'd rather see green kanji crawling up the walls than a bunch of stupid flickery lights...