Here you are saying that Cyan and blue are response of XT signal at one ear or both? Sorry for being so stupid not quite understanding them.
Thank you for your kind words. I first watched Prof. Edgar's (bacch) Q&A page and videos and thought, "This is simpler than I thought." The principle is very clear in the paper.
But Bacch's advantage was... real-time DSP processing.
It's crazy to do this manually.... =(
I'm going to be a little quiet on this one. Because I can't be sure.(You know why, right?)
My guess is that the shape of that signal is the response of the opposite ear. More precisely, the appearance of the
Shadow or shade on my face. (Actual crosstalk excluding ear resonance and room response). (It's one side each, because the left ear of the left speaker and the right ear of the right speaker are not a problem.)
In fact, When i start my process, I was very confused when I first thought about this too. "What's That? What i have to do? What's Actual Crosstalk?"
As you know, I have 4ch impulse. That means, At any time, I can duplicate, edit, or even delete each channel.
So the first thing I tried was to delete LR and RL altogether. Only LL,LR.
This will make it sound binaural. All the information about crosstalk is gone (kind of like Van Gogh, but with one ear missing).
But it didn't sound like a speaker at all, and all clues to ITDs and ILDs were gone. (Like Headphone).
So I thought. What does Crosstalk look like that we're trying to get rid of, that we're trying to improve?
I was asked myself, "What is the actual shape" of a form that is "actually" removed from XTC tasks like bacch or Race.
And that's that graph.
The left ear listens when sound comes from the left speaker. Our right ear listens, too, because it has to.
So what's left when you take away the response from each ear, the response from the room, these commonalities?
That's why the violet and gold appear as straight lines, because you're listening to the left speaker in your left ear.
And my right ear has to go through everything that's going on in my ear and in the room to get there from the left speaker.
And if I equalize the common to zero (delete it), all that's left is the shadow of my face at the end.
My wording and approach may be wrong. But the process was followed by looking at the known papers of RACE and BACCH, and the purpose of reverse phase cancellation was the same, and I started to see significant XTC, both audibly and measurably.
So I still need to improve it further, but it doesn't seem to be completely wrong BACCH-like.
⬈⬈⬈I think my translator is very strange. It's very misleading. My workflow needs improvement, but I don't think I took the completely wrong path because the theory of bacch or race seems to be implemented to some extent.