I hope the prices of your capsules haven't "extra" doubled for me after our discussion....Otherwise I'd better shut my mouth next time...I value your work more than anything
Peace and cheers Ari
No!! Believe me, i understand! It's nonsensical and it made me do a double-take when I realized they were doing it too, especially since they give such good value and they're one of the only places to get this stuff without really expensive EU import fees. This conversation had me re-evaluating the recordings, but my conclusion is the same. They have a position in the market where competition is difficult and I feel they're exploiting that a bit, which makes me uncomfortable, but it also leads to pretty strong opinions about them, since a lot of people have some positive experiences too.
Here's something a little more direct and maybe more in your wheelhouse. I suspect now that I may have been oversimplifying rather than overcomplicating my argument. I apologize for this, and for suggesting you didn't understand. Here's a spectrogram of the delta between the two files, gain-correlated and sample-correlated to the best of my ability:
Notice that the resulting canceled signal is too "full-band." There is
not the expected amount of the distance-based correlation of low frequency sounds you mentioned. If this were real, you'd see greater falloff from high to low frequency or inverse from close to far sources as the correlation between the two tracks due to the distance between the two mics. There's also no meaningful pattern of periodic interference over frequency either to suggest that they're in slightly different places in a room with walls (even a treated one).
Now, what IS there, are clear differences from dynamics processing. There are several of these, but let's go straight to the most obvious one. See that full-band haze, around each of the parts with loud signal but not in the breaks in between? That's the room noise being pumped by the compressor. That, alone, I think says everything. If you listen with headphones, you can actually hear this.
To get further into it, the differences in signal intensity are also too fast and too sudden to be caused simply by a singer varying position between two microphones, and they have a telltale pattern of transient, then gain reduction (which remember, you hear and see twice, since we're looking at a delta and it goes out, in and out of cancellation with the dry signal during one cycle of GR, since the dry signal gain is matched to the average gain, which is the center of the gain reduction curve. If I matched it to the input gain, this would be more obvious, and I will try) and then release. These take the same amount of time after every transient.
Now, we can take a look at the spectrum too, and the difference curve between both spectrums:
First, these spectrums are also way too similar. This is (supposedly) a comparison between a real C12 with a real, chambered CK-12 capsule and a clone with an edge-terminated K67. And yet, the responses of these two capsules, with two totally different anatomies, are this similar within several feet of distance to the source. That's impossible. Probably the most impossible part of all of this. Because of the chambered construction of the CK12, the lobe of pickup surrounding the capsule is just too differently shaped.
Note the very smooth difference curve. The difference curve between the two is NOT smoothed. This is the response curve of a filter or process, definitely. Either EQ, or compressor FR (from being multiband, or it just naturally varies by frequency)
There are also waveform-based arguments that I could make. Here are the 2 waveforms back to back:
Take a close look at the difference in gain. There is a combination of compressor-based and EQ-based difference here. Notice that the gain changes and evens out for the loud parts, but not the soft parts. That is to say, there is no gain reduction below the threshold of the compressor. To some degree, the visible difference is also due to EQ, since the LF stuff has been cut slightly (but not a ton) leaving the HF stuff like esses (which are identical between both recordings, to note) at the same gain. These two waveforms are also just plain too similar to be two recordings of the same event.
Let's compare all these aspects with an actual shootout between an actual u87 and a u87 clone done at Glasshouse Studios. Note that these are orders of magnitude more similar than a real C12 and a C12 clone comparison
would be, given that the u87 and clone have the same basic capsule anatomy.
View attachment real u87 cut.wav
View attachment clone u87 cut.wav
Delta:
View attachment real u87 diff.wav
In order:
Spectrogram of the difference!
Visible periods of cancellation over frequency, no odd noise behavior, no bizarre (and audible) changes in dynamics from the gain reduction pulling the signal in and out of cancellation. Perfect. There's also an audible true stereo field of the room if you pan L and R, which is not true of the M&M comparison.
Spectrum!
Differences due to circuit variation, position and capsule are visible. Lots of little resonant differences. Even though it's a relatively flat curve, it's not smooth, and you would not expect it to be.
Waveforms!
Natural variations in the waveform due to the differences in position and FR.
The M&M comparisons are definitely fake.
