Engineers Use Electrical Hum To Fight Crime 167
Hugh Pickens writes writes "A suspected terrorist has been taped planning a deadly attack and the police want to use this evidence in court, or someone has been captured on CCTV threatening an assault. Increasingly, recordings like these are playing a role in criminal investigations, but how can the police be sure that the audio evidence is genuine and has not been cleverly edited? Now Rebecca Morelle writes on BBC that a technique known as Electric Network Frequency (ENF) analysis is helping forensic scientists separate genuine, unedited recordings from those that have been tampered with and the technique has already been used in court. Any digital recording made near an electrical power source will pick up noise from the grid that will be embedded throughout the audio. This buzz is an annoyance for sound engineers trying to make the highest quality recordings, but for forensic experts, it has turned out to be an invaluable tool in the fight against crime. Due to unbalances in production and consumption of electrical energy, the ENF is known to fluctuate slightly over time rather than being stuck to its exact set point so if you look at the frequency over time, you can see minute fluctuations and the pattern of these random changes in frequency is unique over time providing a digital watermark on every recording. Forensic Scientist Philip Harrison has been logging the hum on the national grid in the UK for several years. 'Even if [the hum] is picked up at a very low level that you cannot hear, we can extract this information,' says Dr. Harrison. 'If we have we can extract [the hum] and compare it with the database, if it is a continuous recording, it will all match up nicely.'"
doesn't work in most cases (Score:4, Interesting)
Workaround (Score:5, Interesting)
So, if I apply a tracking notch to the doctored recordings to remove the original hum and its harmonics, and then superimpose data from the database, I'm all set? Well, thank you!
Don't see it myself (Score:5, Interesting)
Granted, to catch bad fakes and casual deception, this probably works
But what's stopping someone now knowing this, and supplanting their own "hum" back into the right place from any recording make anywhere during the time they want to "pretend" to be recording in?
What's actually stopping them ripping out the hum entirely and replacing it with the hum of any arbitrary period of time? If it's in every recording, and you can spot it in, say, CCTV recordings (that use quite primitive recording equipment and compression methods), then it's also incredibly easy to detect and "fake" yourself.
Also, I don't believe it's as reliable as they make out. We'll find out as more cases use it and it will have to be challenged at some point but even speed-cameras that weren't entirely accurate got a lot of people in trouble and then had to be rescinded years later.
And, totally off-topic, it reminds me of my German teacher in school (I'm English, and we were taught German as an optional foreign language). He basically begged me to take his class instead of Double Science, many moons ago, and even called me a "rat" at a parent's evening in front of my dad when I said I wasn't going to study German any more. Turns out he was kinda hoping I'd take German so he could up the average result a bit.
Two years later, he was sacked because the audio recordings of oral work that he sent off for the final exam had "clicks" on the tape (yes, tape!) in between every question where he'd obviously paused it and briefed the children on the right answer.
Wonder how they submit oral coursework now, with MP3's and things? It would be the work of a second to get a perfectly smooth recording of the same thing happening nowadays.
Re:Workaround (Score:5, Interesting)
That was my first thought.
Though you don't even need access to the database. If you know in advance that you're going to need to doctor recordings, you could just record continuously with a modest amount of data, and filter out all but the hum using the inverse of the notch filter.
You can then so more or less the same process without needing a big database.
I strongly suspect that you wouldn't need much more than sox and octave (using fft/ifft) to do this.
If they start believing this, then you could set someone up by making it appear that a recording has happened at a particular time, when it has not.
Re:Police (Score:4, Interesting)
more like how can the defense lawyers know the police rent railroading them!
Maybe it's just my bias from having worked with a lot of forensic analysts, but usually it seems that it's the prosecutors trying to railroad the police... not into fabricating evidence, per se, but "looking real hard" for "anything they can come up with". The police don't much like it, unless it's a "known bad guy" they're trying to pin *something* on, because searching for trivial things to make the DA look good wastes time that they don't have (not with one-year case backlogs).
or to be fair,m how can the police prove a tape is genuine when the defense lawyer throws up somew fud
This is the most common problem, but "throwing up some FUD" is called "providing a rigorous defense". After all, if you don't periodically make the police prove they didn't fabricate evidence, then there's less incentive to not fabricate evidence.
Re:Uniformity across the country (Score:2, Interesting)
The short answer is: yes, at least under steady-state conditions. Large disturbances (losing a generator, for example) create electromechanical transients that propagate throughout the grid at relatively slow speeds (1000 miles/s).
My research group studies these things - you can learn more here: http://en.wikipedia.org/wiki/FNET
We also have some people working on the ENF authentication technique.
Frequency with Zero crossings? (Score:2, Interesting)