Hummm.....

This seems to have missed on some useful comment:

http://www.theregister.co.uk/2010/06/01/enf_met_police/

The Police have been tracking changes in the exact frequency of the 
mains hum for the last five years - and they are now using That Record 
to validate recordings. Apparently it's the same nationwide.

Most sound (and other) recordings will have a bit of mains hum in the 
background - it's pretty well unavoidable. It's detectable at very low 
levels using Fourier transform technology.

Crystal frequency sources, as used in digital recorders, are stable 
enough that the pattern of changes in mains hum in a recording made 
using them is supposedly detectable, and can later be compared to The 
Record.

Doesn't work with tapes apparently though, the media timing isn't 
accurate enough.

It's supposed to be able to detect recordings made at different times 
than when they are supposed to have been made, and also faked 
cut-and-splice/edited recordings.

Of course this is easily fakeable - if you know about it. And have a 
copy of The Record.

However all may not be as pretty as it seems : according to El Reg
"The police force that ran the investigation this week declined to name 
the murderer in response to requests from The Register, citing 

Re: Hummm.....

Peter Fairbrother <zenadsl6186@...> writes:

> and they are now using That Record to validate recordings.

This is, of course, impossible -- you can prove a recording has been
faked; you can't prove it's real -- but I'd guess that's just sloppy
reporting on the Register's part. Tracking mains hum (or TV line
frequency whistle) is a standard technique for fixing speed variations
in old recordings, so this sounds like a perfectly sensible approach.

> Of course this is easily fakeable - if you know about it. And have a
> copy of The Record.

The National Grid frequency is published in real time, so you don't even
need to measure it yourself:

http://www.nationalgrid.com/uk/Electricity/Data/Realtime/Frequency/Freq60.htm

This is one of those neat systems that's not very well publicised. The
frequency doesn't just vary according to load, as the Register article
suggests; it's deliberately altered in advance of expected high loads,
and various large consumers of electricity have agreements with their
suppliers to go offline or start generators feeding back into the grid
if the frequency drops below a certain point.

--

-- 
Adam Sampson                                         <http://offog.org/>

Re: Hummm.....

On 3 June 2010 18:53, Peter Fairbrother <zenadsl6186@...> wrote:
> This seems to have missed on some useful comment:
>
> http://www.theregister.co.uk/2010/06/01/enf_met_police/
>
>
> The Police have been tracking changes in the exact frequency of the mains
> hum for the last five years - and they are now using That Record to validate
> recordings. Apparently it's the same nationwide.
>
> Most sound (and other) recordings will have a bit of mains hum in the
> background - it's pretty well unavoidable. It's detectable at very low
> levels using Fourier transform technology.

This is a paper (pdf) by Catlin Gregoras
http://www.diamondcut.com/Downloads/AppNote4DiamondCut.pdf who is
referenced in the Reg article.

He's basically using a 49-51 Hz band pass filter to isolate the hum
and them pattern matching against a database.

It would be trivial to remove the hum and replace it with an
arbitrarily different hum. What would be less trivial would be to do
that without leaving evidence that  the recording had been tampered
with (digital filters can, for example, introduce detectable phase
changes).

Looks like you this could be used as evidence that a recording has
been doctored but not the reverse.

Re: Hummm.....

On Thu, Jun 3, 2010 at 9:43 PM, Adam Sampson <A.T.Sampson@...> wrote:
> Peter Fairbrother <zenadsl6186@...> writes:
>
>> and they are now using That Record to validate recordings.
>
> This is, of course, impossible -- you can prove a recording has been
> faked; you can't prove it's real -- but I'd guess that's just sloppy
> reporting on the Register's part. Tracking mains hum (or TV line
> frequency whistle) is a standard technique for fixing speed variations
> in old recordings, so this sounds like a perfectly sensible approach.
>
>> Of course this is easily fakeable - if you know about it. And have a
>> copy of The Record.
>
> The National Grid frequency is published in real time, so you don't even
> need to measure it yourself:
>
> http://www.nationalgrid.com/uk/Electricity/Data/Realtime/Frequency/Freq60.htm
>
> This is one of those neat systems that's not very well publicised. The
> frequency doesn't just vary according to load, as the Register article
> suggests; it's deliberately altered in advance of expected high loads,
> and various large consumers of electricity have agreements with their
> suppliers to go offline or start generators feeding back into the grid
> if the frequency drops below a certain point.

Is there a raw data feed available somewhere too? The graphical view
is very nice and all but not so easy to archive and query.

cheers,

RE: Hummm.....

> On Thu, Jun 3, 2010 at 9:43 PM, Adam Sampson <A.T.Sampson@...>
> wrote:
> > Peter Fairbrother <zenadsl6186@...> writes:
> >
> >> and they are now using That Record to validate recordings.
> >
> > This is, of course, impossible -- you can prove a recording has been
> > faked; you can't prove it's real -- but I'd guess that's just sloppy
> > reporting on the Register's part. Tracking mains hum (or TV line
> > frequency whistle) is a standard technique for fixing speed
> variations
> > in old recordings, so this sounds like a perfectly sensible approach.

All the mains-powered kit surrounding me at the moment is powered from a
UPS.

Analysis of any recordings made in the vicinity would be interesting...

James Firth

Re: Hummm.....

Re: Hummm.....

Re: Hummm.....

> All the mains-powered kit surrounding me at the moment is powered from a UPS

... which I would kind of expect to be phase locked to to mains ...

"serious" ones have a by-pass to raw mains in case of failure of the UPS, 
which they want to be able to use without have a phase change as they cut over.

(and I suspect some "toy" ones use the mains ASIS, and only start generating 
when it fails)

Re: Hummm.....

On 04/06/10 10:29, Dan Brickley wrote:

> Is there a raw data feed available somewhere too? The graphical view
> is very nice and all but not so easy to archive and query.

The site is fiddly, but I found it:

<http://www.bmreports.com/bsp/additional/saveoutput.php?element=rollingfrequencyhistoric&output=XML>

<http://www.bmreports.com/bsp/additional/saveoutput.php?element=rollingfrequencyhistoric&output=CSV>

Go to:

<http://www.bmreports.com/bsp/bsp_home.htm>

Click on relevant historic / current button. Note that you need to
unblock flash.

Damian

Re: Hummm.....

John Wilson wrote:
> On 3 June 2010 18:53, Peter Fairbrother <zenadsl6186@...> wrote:
>> This seems to have missed on some useful comment:
>>
>> http://www.theregister.co.uk/2010/06/01/enf_met_police/
>>
>>
>> The Police have been tracking changes in the exact frequency of the mains
>> hum for the last five years - and they are now using That Record to validate
>> recordings. Apparently it's the same nationwide.
>>
>> Most sound (and other) recordings will have a bit of mains hum in the
>> background - it's pretty well unavoidable. It's detectable at very low
>> levels using Fourier transform technology.
> 
> 
> This is a paper (pdf) by Catlin Gregoras
> http://www.diamondcut.com/Downloads/AppNote4DiamondCut.pdf who is
> referenced in the Reg article.
> 
> He's basically using a 49-51 Hz band pass filter to isolate the hum
> and them pattern matching against a database.
> 
> It would be trivial to remove the hum and replace it with an
> arbitrarily different hum. What would be less trivial would be to do
> that without leaving evidence that  the recording had been tampered
> with (digital filters can, for example, introduce detectable phase
> changes).

Historic filters can avoid that problem.

> 
> Looks like you this could be used as evidence that a recording has
> been doctored but not the reverse.

It was used in the reverse sense, apparently, in the murder case in 
question. Though why the Police won't name the case I don't know.

> 
> How much would the prosecution have to disclose if they used this in
> evidence? Would they have to disclose the code used to do the
> analysis?

Yes, most likely - but it may have been a case of "we can present this 
evidence if you continue to claim the recording was doctored".

-- Peter Fairbrother