Mickey Kantor did NOT utter a racial epithet - forensic proof (large picture)

anigbrowl

(1000+ posts) Send PM | Profile | Ignore

Fri May-02-08 05:58 PM
Original message

Mickey Kantor did NOT utter a racial epithet - forensic proof (large picture)

Edited on Fri May-02-08 05:58 PM by anigbrowl

There's been a lot of controversy about this today. Here I offer proof that he did not say the 'White nigger' remark attirubted to him. I am sorry about repeating the offending word, I do so only for documentation purposes. I don't believe in writing things like f**k and so on - we are (mostly) adults here.

First, my credentials: I am a professional audio engineer who works in film and TV. I make my living by recording sound on location and editing it in the studio. My 'everyday' microphone is a Schoeps CMIT 5U, which you can pick up for a mere $2000 or so. My total equipment setup is worth close to $20k - I say this not to impress you with my spending ability, but to point out that I am heavily invested in this profession (or I would have a lot more in the bank...).

I downloaded a clip which another member kindly uploaded. Yeah, I hear the offending n-word remarks. Eeeew. But what does that mean? NOTHING. It's easy to edit that stuff in, and very few people are actually as good at lip reading as they would like to believe. I know this because I have faked dialog and changed it in the editing room many many times when the original take did not work for some reason.

Furthermore, the people jumping up and down and insisting that they know what they saw and heard are deluded. None of them seem to have taken on board the fact that video on youtube runs at a different frame rate in order to minimize bandwidth. The original video for this film runs at 29.97 frames per second (I know that without having access to it because it's a SMPTE standard which has governed the manufacture of video cameras since the early days of TV). Youtube videos are generally squeezed down to between 15 and 20 FPS, but nobody thought of that before deciding that they were skilled in forensic audio analysis.

Well, none of this tells us whether it was doctored or not. But what does tell us is taking the audio into an editor and viewing it as a spectrogram.

The graph shows you the audio split into different frequency bands. From this we can deduce things about the provenance of the audio, including whether it has been modified. I am using a nicer spectrogram, from Adobe Audition 3, which allows me to zoom in on different parts of the audio and so forth.

In the audio I downloaded, there are two key characteristics. One is that the frequencies are rolled off above 20Khz. This was almost done by a low-pass filter in the recording mechanism of the camera, which is a standard part of the design. Secondly, there is a hole centered around 11Khz, above which the fequency spectrum is mirrored. The mirroring is due to a phenomenon called aliasing which occurs following certain signal processing operations applied to digital audio. The most likely explanation is that the camera was set to record audio in relatively low quality, topping out around 11KHz. Again, this is a standard thing on video cameras because it means you can squeeze two or four tracks of digital audio recording into a tape instead of just one or two. It's inadequate for a Hollywood movie, but quite OK for a documentary. Most video cameras have an option to switch between full and limited audio sample rates, and 11KHz is quite adequate for documentary sound pickup.

So, I know just from looking at the spectrum how this sound was recorded and I know that it is consistent with in-camera audio recorded at a low sample rate.

BUT when I look at the section where the whispered words 'like to be a white nigger' are heard, it's different. I can see spectral information extending upward in a smooth line all the way up to 22.5 KHz every time there is a plosive transient (go look up what those words mean). This is completely inconsistent with the rest of the audio on the recording, and shows that that segment has not been through the same low-pass filter as the rest of the recorded audio - ergo, it was dubbed.

I would be perfectly comfortable going into court and demonstrating step by step how such a recording could be assembled and still sound real - but the spectral information in the recording makes it abundantly clear that other audio has been dubbed on top, just as there are clued for the experienced eye when looking at a picture that has been modified in Photoshop.

It is actually possible to edit out this spectral information, but few sound editing packages offer this kind of functionality (ie applying edits to only a partion of the frequency spectrum). Mine does; if I had edited this, I could certainly have removed the offending 'fingerprints'. But whoever put the original thing together either did not have access to or did not use the spectral display, and thus left proof of their handiwork all over the file. Ooops.

Bottom line: the phrase 'white nigger' is FAKE and was added to the original recording.

What do I conclude from that? Nothing. Might have been a bitter Obama supporter. Might have been a prankster who wanted to troll. Might have been a false-flag operation by someone on the Clinton or GOP side who decided to offer the Obama people a live hand grenade dressed up as a peach. I have no idea and there is no way to tell. Conspiracy theories are nothing more than theories. Without evidence, you would be foolish to draw any conclusions.

Semi-related: I don't know what he meant about the 'shit' remark ('Look at Indiana - 42-40. Those people are shit--excuse me..'). He might mean the voters, the state campaign team, the pollsters, the Bush campaign team, or anything. I really don't care.

Send me a private message if you want to pass this info on and get my contact details. I already sent this info to a newsweek blogger. I am happy to document it in excruciating technical detail if required.