On Fri, Apr 20, 2012 at 9:27 AM, Michael Dalton <>w=
rote:
> Is there any defense against "fabricating data," i.e., slicing unrelated
> sounds together to make them appear to be related? The only thing that I
> have noted so far is that on a few recordings there are background sounds
> such as a Cardinal chirping in the background that are continuous through
> my samples. On the other hand, it is possible to put in such sounds and
> then edit them closely to make them appear to be part of the clip.
I'd say the chance of a successful fabrication is 'proportional to the care
of its creation. :)
Contrariwise, there may be fabrications that are effectively impossible
give current technology to debunk -- but it would take extreme care and
attention to detail probably all out of proportion fo the benefit of so
doing. :)
Assuming recordings nominally made continuously from a static postion,
If I were personally asked to verify the veracity of a recording, I'd start
with easy things first -- scanning the waveform for truncation and obvious
variations in background noise. Then I'd look closer at the ambiance,
spectrographically. Matching its perceived level is one thing; hiding
changes from different levels of mix is another.
You could also look attempt to abstract out the microphone array used
(which in your example should be documented anyway) and see if it's common
and consistent across all sound events. Variations in timing or
frequency-specific timing, if you could distinguish them from simple stereo
positioning and local "real" coloration, could be a clue. Blending of
samples from multiple signal paths (including original sample rate) might
be detectable, I suppose, especially if different mics/pres/rates were used
with different roll off characteristics at the top and bottom.
That's just looking at the technical aspects -- someone with appropriate
domain knowledge could probably also discern anomalies in the biophany that
would at least raise questions...
But that's a lot of work to go to -- on both ends! :)
aaron
--
quietamerican.org
oneminutevacation.org
83% happy
9% disgusted
6% fearful
2% angry
|