Here is the message I sent on Greg Nichols’ Youtube site some days ago, with whom I have been in a lot of dialogue over the last two months.
INCREDIBLE BUT TRUE, SHOT 10 WAS FIRED 16 M NORTH OF TRUMP!
I came to the conclusion that obtaining Boom times by calculating tdiff has two major drawbacks: you have to assume the location of the shot and guess the ammunition used.
So I tried to use only the raw data without doing any calculations. So I eliminated Ross, DJStew, Cruiser and, initially, Podium from the simulation. I removed the RAV recording because the time difference between shot 9 and shot 10 is different between the three broadcasts with cuts in the first (18 ms) and the third (7 ms) making it impossible to know the real time difference. I replaced Podium with a recording made nearby: source 1 in Googledrive by bruin_jim. I also used “don’t run”, NTD and “Audio source 4”.
I have the following Boom times:
Source 1: 16.000 s, TMX: 16.272 s, Dontrun: 16.182 s, Source 4: 16.373 s, NTD: 16.076 s.
The positions of these sources are as follows:
Source 1: (586740, 4523400)
TMX: (586730, 4523516)
Dontrun: (586654, 4523415)
Source 4: (586684, 4523545)
NTD: (586691, 4523403.5)
The podium is at: (586726.6, 4523394.3) and the shooter at: (586767.7, 4523529.1)
I get such a perfect convergence that I was surprised myself at (586728, 4523410) or 16 m north of Trump’s position. It seems incredible but the convergence is so good and independent of unknown data that I think it is the right solution. It is so close to the podium microphone that it explains the very loud sound that you took for a Crack when it is a Boom that saturates the recording.
Finally I reincorporated Podium in the simulation but it takes a time difference of 10.030 s between shots 9 and 10 to have the convergence at the same point, that is to say a Boom time of 16.010 s.
I used the same method for shot 9 and I have a very good convergence at (586754, 4523422)
I leave it to you to redo these simulations to confirm my result. And to find the author of the shot! Maybe one guy from HawkEye team you spoke about.
In my simulation, I therefore have 6 sources and 15 hyperbolas all converging at the same point: very impressive!