O RLY?
By a strange coincidence I've lately become interested in learning about the characteristics of authentic demolition blasts as recorded in internet videos. To that end I've been garnering soundtracks of real demolitions from YouTube, capturing the audio with Wiretap Pro, a program which intercepts the data going to the computer's sound card and records it as an audio file.
Ultimately I hope to be able to provide some statistics on such parameters of interest as risetime and spectral distribution. However, even with the relatively crude tools I have available at workplace #1 we can examine a few of those soundtracks and see if we can determine just which frequencies you would have to filter out to suppress the sound of chaped charges going off.
Let's start with the demolition of the Landmark Tower in Texas. I selected the first four blasts (selection is shown in the waveform display below)
[qimg]http://www.internationalskeptics.com/forums/imagehosting/3784b841a85bd149.jpg[/qimg]
and did an FFT spectrum analysis of the selected portion:
[qimg]http://www.internationalskeptics.com/forums/imagehosting/3784b841ada2e751.jpg[/qimg]
The horizontal axis is logarithmic with the major divisions being 1 kHz apart. The vertical axis is linear in decibels, 6 dB/div.
You'll notice that while there is quite a bit of bass, the spectrum is nearly flat out past 3 kHz and continuous. Now let's look at a time-versus-spectral density display for the same selection:
[qimg]http://www.internationalskeptics.com/forums/imagehosting/3784b841a85a8741.jpg[/qimg]
Horizontal axis is time, linear, 1 second/div. Vertical axis is frequency, linear, 1 kHz/div. Intensity within a frequency bin is denoted by color, with brighter colors indicating higher amplitude.
The individual blasts show up nicely in this display; they're hard to distinguish in the waveform display because each succeeding blast goes off before the sound of the previous one has finished bucketing around the gutted building. You'll notice that each shows a continuous spectrum with considerable energy out beyond 5 kHz. This is consistent with the audible fact that instead of going "BOOM" or "Whoom", the explosions go
[qimg]http://www.internationalskeptics.com/forums/imagehosting/3784b842798d0c77.jpg[/qimg]
BAM! BAM! BAM-BAM-BAM!
So, if you want to use filtering to remove the sound of the explosions, you would have to take out everything between DC and 6 kHz or so. The technical term for doing this is "making your recording utterly unintelligible and useless".
Now, let's look at a couple more demolition soundtracks. The Everglades Hotel:
[qimg]http://www.internationalskeptics.com/forums/imagehosting/3784b8419e88978b.jpg[/qimg]
spectrum
[qimg]http://www.internationalskeptics.com/forums/imagehosting/3784b842a72d4f7b.jpg[/qimg]
spectrogram
The Kingdome in Seattle:
[qimg]http://www.internationalskeptics.com/forums/imagehosting/3784b8419e8c0026.jpg[/qimg]
spectrum
[qimg]http://www.internationalskeptics.com/forums/imagehosting/3784b842a72e5add.jpg[/qimg]
spectrogram
They both have something in common with the Landmark demolition: the explosions produce continuous spectra with plenty of energy up into the midrange. The notion that frequency-selective filtering could remove the blasts from the recording without introducing painfully obvious artifacts is nothing short of laughable.
I've experimented with using some digital single-ended noise reduction techniques to try to remove the blast sounds. Multiband noise gating had no effect because, as I tried to explain to femr2 above, the blasts have a higher amplitude than the background sound and consequently just open the gates wide. Using spectral subtraction with the first blast as a sample of the "noise" to be reduced just turned the entire track into something that sounded like R2D2 inciting a riot in an aviary.
Which is more probable- that the eevil gummint has some magical technique for removing the sound of explosions from audio recordings without leaving a trace and has done this with every WTC collapse video floating around the 'net, or that people with an axe to grind or a twisted sense of humor have occasionally added phony sound tracks to existing videos and uploaded them to YouTube just to wind-up rubes who, in the immortal words of Honest John Barlow "won't know enough to do any smart checking"?