backtop


Print 27 comment(s) - last by drycrust3.. on Dec 21 at 3:24 PM

UK police are putting tactic to use to fight crime

Romanian audio specialist Dr. Catalan Grigoras, now director of the National Center for Media Forensics at the University of Colorado, Denver, made an intriguing discovery about a decade ago. The ubiquitous hum of modern society follows a unique pattern that allows many recordings to be validated.  Now police in the United Kingdom have begun to use the tactic to verify evidence in important court cases.

I. Industry's Silent Song

Recordings traditionally have been a highly unreliable form of evidence, given that they could easily be cleverly staged or tampered with.

That's where the hum comes in.  Electrical sources such as light poles and power outlets emit a near imperceptible hum.  While centered around the frequency of the alternating current (50 Hz in the UK), the hum dips and rises by a few thousandths of a hertz over time.  The frequency drops when demand outpaces supply, and rises when supply outpaces demand.

Given a long enough window, this pattern of rising and falling frequencies is virtually unique, as Dr. Grigoras found.

Power Grid grid
[Image Source: UCSC Silicon Valley Extension]

But by using a technique called Electric Network Frequency (ENF) analysis, law enforcement can store the pattern of the hum for a particular grid in a database.  The Metropolitan Police lab has been compiling such a database in recent years, as has JP French Associates -- another UK forensics lab.

Comments JP French's Dr. Phillip Harrison to BBC News, "We can extract [the hum from a recording] and compare it with the database - if it is a continuous recording, it will all match up nicely.  If we've got some breaks in the recording, if it's been stopped and started, the profiles won't match or there will be a section missing. Or if it has come from two different recordings looking as if it is one, we'll have two different profiles within that one recording."

II. A New Time Stamp, but Could it be Gamed?

A trio of London gangsters -- Hume Bent, Carlos Moncrieffe and Christopher McKenzie -- recently saw their defense against London Metropolitan Police charges of gun dealing fall apart thanks to ENF.  Dr. Alan Cooper, a Met Police ENF expert, validated police recordings of weapons deals using the grid buzz, scientifically damaging the defense's claim that the recordings were tampered with.

London gang suspects
London Gang guns
A trio of gun dealers were sentenced with the help of ENF verification of police evidence.
[Image Souce: PA]

The trio was founded guilty and sentenced to prison for a total of 33 years.

It seems appropriate the novel forensics method has been pioneered in the birthplace of fiction's Sherlock Holmes.  But in years ahead, some questions about ENF remain unanswered.  For example, while individuals would be unlikely to be able fake the ENF hum, it might be feasible, albeit extremely difficult, for a police force to filter out the hum in a recording and dub in a hum at the time they wish to make the recording appear from, given that they have access to the entire database of recordings.

It might be even possible for a citizen skilled in audio recording to carry out such a feat.  Thus the technique may lay to rest questions of cruder tampering, but may still have flaws of its own.  For that reason, in time it will probably be used as a piece of a richer evidence puzzle, also composed of other circumstantial clues like cell phone tower records or surveillance footage.

Source: BBC



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Hmmmmm...
By drycrust3 on 12/14/2012 2:53:48 PM , Rating: 2
quote:
a Met Police ENF expert, validated police recordings of weapons deals using the grid buzz, scientifically damaging the defense's claim that the recordings were tampered with.

quote:
For example, while individuals would be unlikely to be able fake the ENF hum, it might be feasible, albeit extremely difficult, for a police force to filter out the hum in a recording and dub in a hum at the time they wish to make the recording appear from, given that they have access to the entire database of recordings.

Actually, it would be extremely easy. Say you made an illegal "wire tap" recording of Jason talking to Tiffany over the office phone system talking about they think the 20th of December predictions are nonsense, how much mains hum is there on the phone network? Next to nothing! Ok, so what about the recording device? If it's battery powered "mp3" type recorder then again hum wouldn't be perceptible if it was running just on the batteries. There might be some if the recorder was running off a charger. So say it the recording was just done with batteries, we now have a recording with no hum on it. Now we want to edit the conversation so it sounds like Jason and Tiffany believe the 20th December is the end of the world ... queue in a basic PC audio editing software ... change the conversation ... add in some hum ... voila! Job done!
The reason you get hum is because of low quality filtering on your power supply, phone systems use extremely high quality filtering, hence no perceptible hum.
To me, the Met should be explaining how the hum got onto the conversation.




RE: Hmmmmm...
By Jeremy87 on 12/14/2012 3:25:29 PM , Rating: 2
And you'd think that lossy encoding like mp3 would filter out things we can't hear well, like this hum.


RE: Hmmmmm...
By drycrust3 on 12/14/2012 4:06:52 PM , Rating: 2
Pick up any phone receiver, wait for dial tone, press a digit to cancel out the dial tone, and what can you hear? Hum? I think not!
In New Zealand, where I live, there is no perceptible hum on our phone lines, and our system was originally based on the UK system, so I would expect to get the same result there. It might be different in America, but I be surprised if that was so. Go to a phone and try it out.
My experience is hum is apparent on cheap (and not so cheap) speaker systems.
If there is no hum, then you don't need to filter it out.


RE: Hmmmmm...
By JediJeb on 12/14/2012 5:09:43 PM , Rating: 2
Plug in an old analog phone and you will probably hear much more hum than you think you would. I have found that using a modern cordless phone will filter out a lot of the hum, crackle and other noise from a phone line. When the local phone company wants you to do a test for line damage or noise they always ask if you have a regular phone to connect instead of a cordless version. Wiretaps would be placed at the network interface box which would be before the filters and would be able to pick up the hum.

Also if there are fluorescent lights in the office the hum can come from them and even appear on the recording done on the battery operated mp3 player. I think this is probably what they are alluding to as much as that on the phone line.


RE: Hmmmmm...
By MadMan007 on 12/15/2012 4:17:38 AM , Rating: 2
"a near imperceptible hum" This isn't the same as the ground loop hum you hear from speaker systems.


RE: Hmmmmm...
By drycrust3 on 12/15/2012 8:30:49 AM , Rating: 2
True.
The real question is how easy would it be to make recordings of conversations without hum, edit them, and then add the hum.
I said that as far as I could tell it would be very easy to make recordings of phone conversations which didn't have any discernible hum, but one of the other commenters said there would still be hum, it just wasn't audible from a phone, but would be to a normal recording device. The hum being a means of confirming the conversations took place at one time instead of several at different times.
Without knowing the exact nature of what was recorded and how it was recorded we aren't able to know how credible this technology is. The problem here is 3 guys are have been sent to jail for a long time, and no one has had a chance to peer review the technology that sent them there.


RE: Hmmmmm...
By HoosierEngineer5 on 12/15/2012 10:05:19 AM , Rating: 2
Using digital signal processing, it might be possible to pick out hum that is inaudible (taking the discrete Fourier transform of a sufficiently long record).

For digital recordings, sufficient signal or noise would need to be present to get the data samples to change. Unfortunately, as some point out, filtering and nonlinear compression of some codecs can affect the signal and potentially make the detection process suspect. This might force the analysis to use 'judgement' to reconstruct the record, which introduces the possibility of bias.

If an analog recorder (using motors) were used to capture the signal, 'wow and flutter' could render the technique unreliable.


RE: Hmmmmm...
By mmatis on 12/15/2012 9:45:52 AM , Rating: 1
Ah, but they are "Law Enforcement", and do not have to explain ANYTHING! State secrets. Or "we're the police, so we're more honest" even though at least 20% of their sworn testimony and evidence is testilying. And the judges will make sure that nobody who disbelieves those lies is empaneled on a jury.


RE: Hmmmmm...
By drycrust3 on 12/15/2012 8:09:25 PM , Rating: 2
From time to time we hear of various "forensic" methods which were well established and accepted as true, and then suddenly someone discovers that the method is based upon a false assumption, and suddenly there are a ton of people in jail or who have done time and have criminal records who want a retrial.
This is exactly the problem I see here, in that people say you can't fake the hum on the recording, but we don't have proof it can't; and they say the precise frequency of the British 50Hz mains is unique to certain times of the day, but again we don't know for absolute certainty that this is correct. For example a UK UPS generates its own 50 Hz frequency independent of what the mains is doing, thus giving a different impression about the hum than would otherwise be the case, for example the UPS mains could easily slow down when you place a bit too much of a load on it.
Another example is a recorder that uses the mains as a clock. Assuming the clock controlling the recording is precise and unfluctuating would be wrong when in fact it would actually be recording the conversation in almost exact synch with the slowly deviating 50Hz, thus it would again give a false impression regarding the time of day, e.g. make everyone think the recording was made at a time of day when the demand is fairly stable e.g. mid afternoon, early hours of the morning, etc, when it could actually have been made in peak hours; or a woman down the road could be using an arc welder while the man next door is doing some ironing and this causes the local voltage to fluctuate while the mains frequency remains stable, thus the recorder records that bit of conversation as taking place just prior to the evening peak when in fact it might have happened in the middle of the day.
To me, it seems just way too easy to pick holes in the process for this to be trusted, especially when you are looking at the lengths of sentence dished out.


RE: Hmmmmm...
By mmatis on 12/16/2012 11:18:59 AM , Rating: 1
Ah, but the US "Legal" system, at least, would not let a defense attorney "pick holes in the process", for as soon as you are permitted to see the man behind the curtain, the entire charade begins to fall apart. And the "Legal" system will simply not permit that.


"If you look at the last five years, if you look at what major innovations have occurred in computing technology, every single one of them came from AMD. Not a single innovation came from Intel." -- AMD CEO Hector Ruiz in 2007














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki