@pk2019 This is actually a really nice result for the “pulse+echo” correlation. Shot 8 matched shot 4 best as it didn’t have another shot stomping on its echo, but 5, 6, and 7 matched reasonably well, and 1-3 didn’t at all.
I am not sure what happened with just the pulse, maybe it wasn’t long enough to discriminate.
Here are the beginning times I have for all 10 shots and source 4:
OK, the audio forensics guy got back to me. He has been very open and helpful.
I asked about the muting of the higher frequencies. He says that most of the audio data he’s ever asked to look at has gone through lossy perceptual coding (compression), and that he sees the muting effect all the time.
I asked about whether it’s odd for the waveforms to be so similar. He said they are remarkably similar but that it may be due to how high up Crooks was, rather than down on the ground near a bunch of constantly changing stuff like trees. He said that the waveforms looked similar to ones that had been collected at high elevations to emulate an anechoic chamber.
The other factors are things like the barrel heating up and powder loads, but the latter is tightly controlled lately.
He said he was more disturbed about timing differences about the first stream (stream 0):
I did a quick timing measurement and found something off on Channel 0 compared to the other recordings. The timings were the same +/- 1 millisecond for all gunshots in dashcam-1, dashcam-2, and my other recordings. This is a very strong indicator of one shooter at one location. The strange thing is that dashcam-0 has slightly different timings for the first three shots. Typically this happens when the receiver is moving between shots, but that won’t happen if it is a stationary dashcam. We’ve also seen MP4 recordings with perceptual coding that sometimes slip the time during decoding - but I don’t know why that would only happen only on the first three shots on dashcam-0.
I still think this is probably a productive line of inquiry, even if not for the reasons we originally thought.
So far, Matthew Pearson has not answered my questions about the chain of custody of the videos.
Asked and answered many times, and this is the fallacy of just throwing sh1t on the wall hoping something sticks. Sorry but every reflection on those windows has an innocent natural explanation and the same type of reflection is seen on at least 1 other window proving it’s a natural occurrence. The exception is the white light flash, which is not a gunshot. We know this because the window is showing reflections the entire time and at no time was opened/closed or removed/replaced. There was simply no observation nor time for it to be opened/closed or removed/replaced.
Let’s apply some logic, M’kay. The fact we have reflections during the entire time, proves that the window was A) not removed, and B) closed to offer reflections from Stewart’s position.
Go 1 at a time:
“Rope” appearanced. This occurs on both windows 2 and 3, several times, before and after the shooting and corresponds with Stewart in the standing position, and disappearing when he is kneeling. Logic says this is a natural event based on the angle of the camera.
Window 2 and 3 “rope” reflection:
Window 1, 2 and 3 “block” reflections. Window 1 had the lower section with a white block, Window 2 has a white block in the upper section, and Window 3 has an entire white block. These are reflections. Try going outside and looking at windows from different angles, it’s a natural occurrence in nature!
4:59 timestamp, just about 2-3 seconds after the first volley window 3 shows a total reflection proving it is there and closed flat.
Window 3 “rope” and “red” reflection. Based on the angles of the roof channels going to the rights, Stewart films this slightly to the left or SE of the window, and it is flat closed facing the red barns and reflecting the red barns 400 feet away.
1 minute before the shooting, the window is obviously closed.
4:49 timestamped, just 6 seconds before the shooting, the “red reflection” is still observed proving window is still there and closed flat to reflect the red barns.
Sorry to burst your bubble but these are all natural reflections.
The main curiosity is the white light. By a process of elimination, since it does not align even closely with the sound it’s not a gunshot. Furthermore, since no window was removed or opened, it cannot be a gunshot. Furthermore, since the location at about 6’ tall is several feet lower than the bleachers and audience and could not target Trump it was not the gunshots. So it’s something else. Could be a white light flashlight from the interior as someone searched the room seconds after the firing, or a reflection coming from outside such as the headlights of the SUV parked facing it.
As to the balances of the anomalies (like the “smoke” LOL) they look badly photoshopped or simply due to poor video quality.
Let me turn the question on you. What is YOUR explanation why Window 2 has a “rope” in the upper section, or other random patterns, many times including here?
And check out Window 3. It’s obviously reflecting the red barns, and now instead of a “rope” we have a diagonal white block, making a triangle.
This causes me to wonder if some of you have ever experienced the natural outside world…
Hi Bumblebeeez,
OK, most likely you are right.
In seeking a higher resolution of body cam BWC2-122110 I came across this video and simply wanted to share it with you. As you say the sound looks to be offset, you can tell when the cop opens the door, it is not sync with the motion.
Maybe an audio specialist wants to take the time anyways and have a look at the audio? You never know, maybe they did have access to the original file and copied it into this video? If it turns out, not to be the original audio, at least we would know from where this audio has been copied from…
A lot of officiall body cam videos have been released by Butler Township Police. One of big interest for me is of course BWC2-122110. Does anybody know where these originals videos can be downloaded?
The Dropbox provided by Matthew Pearson from Butler PD can be found here: Dropbox
The dashcam video with the shots is the 10th video, and the shots start at about 4:55 into the video.
Brian,
Thanks a lot!
1.8GB high resolution of BWC2-122110 sounds very promising and hopefully can shred more light on the subject.
I’ll get right to work… 
That’s a great catch, thanks for sharing 
Since there seem to be other recording angles, I feel I have to screen another livestream as well…
Hi Roger, I appreciate all the work you are putting in on this.
I have to disagree with the conclusion that the flash we see in window 3 is a muzzle flash, though. Muzzle flashes are not going to last several frames of a video, unless it’s a specialized slow motion video. Those flashes only last a few milliseconds.
It could be a flashlight, or some sort of device that could help synchronize a rifle that’s positioned in the vent in the same room.
I’m personally open to some pretty wild stuff - including self-healing glass, so I’m not going to dismiss anything out of hand. Any videos that were in the authorities’ possession should be considered possibly altered - which makes me wonder why a flash would have been left in “boosted police officer’s” body cam video if it was a genuine clue.
In summary, I’m pretty sure there were two shooters, and the trajectory of the first shot seems to put the first shooter (or their remotely controlled rifle) in AGR building 6. Beyond that, I just don’t know…
Is is also possible Jason Wood’s message at 5:52pm is a reply of GN’s message at 5:45pm, in which he posted the suspicious bike under the tree on the west side?
If JW out of the AGR parking lot east drove left, down Whitestown City Rd towards Evans City Rd and had to stop at the traffic light. If he watched left towards the AGR parking lot west, he might have well recognised that “That bike was not there when I pulled out.”
Unfortunately we can’t see the second sentence in full. Could be something like “Maybe have them [meaning whoever at command] check that car in the other parking lot [refering to his own message from 4:26pm] as well”
I worked for an american one. So far that was the best job in my life. Megacorps are full of people who want to juggle and many times fail. (Someone very professional developper was sad, since all his work within 10 years went to the garbage, as he explained once his frustration.)
The Mazzoni filters introduce delays between 10 and 20 ms. That could be the reason that the 21.360,21.380 s bounds for shot-4 do not include much of the pulse if the filters you applied don’t shift the data by a similar amount.
I already wondered about your 21.360 s offset indication when looking at the extracted shot-4 sample. At first I suspected an error within the av::read_binary_data() function and subsequently a bug within my own code used to sort out the two audio channels from the returned 1-dimensional vector.
edit: typo regarding shot 9 at 20.678 seconds?
We should talk about the spectrum (but maybe the problem is only the lack of my knowledge). How on Earth the lower frequencies can change sample by sample. They have longer time of period.
Yeee, this is the effect was mentioned a few days ago. (I open the newspaper and beleive all.) I look at the spectrum, but actually I don’t know what I can see, how they produced. Maybe FFT. Did they slightly shift the time window?
Brian, what exactly this audio expert meant when he used the word timings?
If he observed a systematic phase shift (~10 ms between d-0 and d-2) then the shots registered by d-1 and d-2 can’t have the same “timings”, because the shots registered by d-1 were timed ~5 ms earlier than those registered by d-2.
So does he equate “timing” with “gun report duration”? Or with delta time between shots?
edit:“by d2” → “by d1” typo corrected
the red barns and their white roof…
yes, I am sure that that bodycam footage will SHRED more light on the subject 
I think he meant the latter, but I’m not sure. I’m hoping he’ll elaborate a bit, but I’m trying to be careful and not irritate him, because he’s a professional who has helped us for free.
It was the ninth shot overall, but the 6th in that fast string after the first three. Best to think of it was the 9th.
I can’t say for sure if it hit Crooks, but it did mark the end of the shooting from AGR6. At a minimum it snapped by alerting him that he was now engaged.
I can do basic stuff with the drone. At what height was it contemplated that Trump’s ear was?
If it is not too much of a hassle, get a zoomed up front picture of building 6 / window 3?
