Hum ho, now where was this recorded?

[Colin McDonald]

This sounds almost like it should be in "Area 51" but I believe it's genuine. I just wondered whether it would really work on professional audio recordings (apart presumably from recordings made on battery powered field equipment in an open location in London eg a park). Some sound guys I know would be offended at the suggestion that there was enough hum present in their recordings to trace the location.

UK police record 7 years of background noise to help fight crime

According to the BBC News BBC News - The hum that helps to fight crime, a police forensics lab has been recording the digital hum of London's electric grid for the past seven years, and this record will allow experts to pinpoint the exact time that any audio recording was made in the city.

Whenever any audio recording is made, it will always pick up background noise, which will include a hum produced by any electrical power lines or appliances in the area.

"The power is sent out over the national grid to factories, shops and of course our homes," said Dr. Alan Cooper, an audio forensics expert with the Metropolitan Police Forensic Audio Laboratory. "Normally this frequency, known as the mains frequency, is about 50 Hz."

This hum isn't constant, though. It changes, ever so slightly, based on the local demands on the power grid, and by recording these changes, the London police have constructed a digital audio 'timeline'.

Now, using a method called Electric Network Frequency (ENF) analysis, police can examine any digital audio recording made in the city over the past seven years (and moving ahead into the future) and compare the background noise from it to this timeline. Matching up the background noise from each will provide a digital time-stamp, and allow them to tell unedited recordings from edited ones. This will prove to be incredibly valuable for authenticating audio evidence presented by both victims and police investigators.

"We can extract [the hum] and compare it with the database," said Philip Harrison, a director of the independent audio forensics laboratory, J. P. French Associates.

"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."

ENF analysis can be used for any power grid and any recording made within that power grid. For the U.K., where the entire country is supplied by one power grid, only one background hum record is needed, but for Canada and the United States, which are supplied by a total of five power grids, this technique becomes a bit more difficult. However, as long as there is a continuous record of the background hum from each grid, the time-stamp of an audio recording can still be found by comparing it to each of them.

Article quoted UK police record 7 years of background noise to help fight crime | Geekquinox - Yahoo! News Canada as I couldn't find the BBC report at first.

[Richard Crowley]

That seems absurd for several reasons.

First, the range of frequency is very very very narrow. I doubt that the power authorities allow the mains frequency to drift more than a tiny fraction of one percent. At least that is the performance over here in the Colonies. There are probably several times per day, for many minutes at a time when the mains frequency is exactly 49.9994Hz.

Second, If you are going to try to establish a time of day based on the exact frequency of the power mains hum, you must assume that there is no time distortion between the recording sample rate and the playback rate. And nobody who has tried to sync a wild audio recording to a video thinks that is possible in the Real World. I am extraordinarily dubious about this.

And if they hope to use non-recurring mains noise, that is so localized that it would work only on the same floor of the same building where they were taking the "reference recordings".

[Greg Miller]

I concur with Richard, for all the reasons given.

You can't tell me that at a given minute on a given day in a given year, the power line frequency was different from some other minute over a seven year span. The same frequency has to occur again and again, at randomly varying times. And you'd need an atomic time base in the recording equipment, to measure this accurately. Given what we know about maintaining sync between various cameras and recorders, most "typical" recording equipment is not nearly accurate enough.

From my experience, I know that analog clocks running synchronous motors often lag behind atomic time, by a second or more, at mid-day in the summer. The power grid runs slow due to the heavy load imposed by air conditioning units. But they catch it up to "real time" by running slightly fast overnight, so clocks are accurate again (for at least one instant) the next day. There is no cumulative error.

Now, as to detecting edits, that's another matter. Yes, if you filter down an audio recording and bandpass only the power line frequency, you should see a sine wave, and if there was an edit, it will show up as a discontinuity in the waveform (unless the edit was done on a digital workstation, taking care to edit at exactly the right place). In fact, IIRC, this was a technique used to show that there had been edits in the "Watergate tapes" many years ago.

[Henry Kenyon]

Sounds like "junk science" to me. Unless there is some kind of subliminal coding system deliberately designed for that purpose, that can be proven in a court of law, I would be surprised (well maybe not), it passes for evidence.

[Steven Digges]

Gentlemen. I am not a conspiracy guy at all. But I can tell you what I heard at a conference last year. The lecture was given to a international group of computer science engineers (I am not one, I was there for a different reason). The lecture was about the science and technology of determining the exact time and location where any piece of video was shot based on examining its digital footprint and many other factors. Every camera, power supply, recording media, and light source has a unique foot print. They like sun light the best. It blew my mind how much they can get from analyzing that, broad geographic location, time of day, possible date of shoot by matching weather conditions to time and area etc. Location discovery by power phase etc, etc. It went on and on. I am not saying they can pin point every time and location of every video yet. I don’t know how close they are. I can tell you there is a lot of money and research going into it. I thought it was all just 1s and 0s, not a chance, that video image it a gold mine of information. It blew my mind.