Very odd

[macc]

Hmmm.

Just had a dnb track to do with a dirty funk break type hi hat running kind of 70% off to the right. The open hat hits were killing me, so I split to MS (Voxengo MSED) and with the UAD precision de-esser, errr, de-essed it. Set it to split mode, hi pass at about 7.5k, doing up to 4dB on the worst ones.

It hit me that suddenly it was on the left. As I played some more, it became apparent that everything above the crossover was going over to the left.

Now for the life of me I can't figure this out. Of course, I need to think about it a bit more first... but ffs I have a physics degree and I can't clock it at this stage of the evening

I swear it is either because it isn't a linear phase crossover, but I thought it was. Or maybe it is cos it isn't but I thought it was. DAGH!

Can anyone help me figure this out? I know I know, but I don't know. Or something


EDIT: 'This filter bank is a three-band implementation of the minimum-phase filter bank used in the Precision Multiband Compressor.' So it is from the MBC, but not the linear phase part. I still can't properly figure it out though :?

[Adam Dempsey]

Do you get the same result processing via same but in L-R mode instead of M-S?

[Darius van H]

You need to test if the MS part is working - sounds to me like it isn't.

[finetuner]

check the panning of the individual M and S channels.

Peter

[macc]

When I bypass the desser - which is just working in normal mode - it all returns to normal. That is to say, straight MS split/recombination works as expected - it is identical to the original. It's only when the de-esser is engaged that everything above the HP cutoff gets flipped


I don't have a de-esser that does L-R, but I could try splitting it that way... Thanks for the thought. thumbsup

[OberHeim-Kenobi]

Originally Posted by macc

When I bypass the desser - which is just working in normal mode - it all returns to normal. That is to say, straight MS split/recombination works as expected - it is identical to the original. It's only when the de-esser is engaged that everything above the HP cutoff gets flipped


I don't have a de-esser that does L-R, but I could try splitting it that way... Thanks for the thought. thumbsup

Hello Macc, I have been doing some stuff with L-R split de-esser and usually got
"more euphonic pleasant results" than chained mode , try to expand and to down level on hi band (UAD Multiband), it will soften hi spike transients in a good way.

Juanjo.

[twwalsh]

the phase might be getting screwed up by the plugin latency.

[macc]

Nope... doing it in cubase (yes, I master in cubase) and it is fully delay compensated.

But that does lead me to consider the phase- it must be the phase effect of the crossover. I just wanted to understand exactly how the interaction works upon recombination (like I said... physics degree).

Is it too much to ask to get some cold hard geek talk around here?!!!11/?!?!1,1

[twwalsh]

delay compensation doesn't always work!!!

it's not fool proof by any stretch.

[macc]

I know

But nevertheless, incorrect delay compensation - bearing in mind the whole plugin is included in that calculation - doesn't explain the cutoff-related behaviour. Not to me anyway... unless I am missing something (else).


Not meaning to be belligerent, just trying to get to the bottom of it

[Adam Dempsey]

Originally Posted by macc

I just wanted to understand exactly how the interaction works upon recombination (like I said... physics degree).

Is it too much to ask to get some cold hard geek talk around here?!!!11/?!?!1,1

Dark matter?
Or maybe E (error)=MS squared?

[macc]

[donsolo]

it sounds like your M-S return matrix got screwed with when you turn on the De-Esser. If you add on the wrong side, you flip the stereo image...

what could cause that? Well if your De-esser is inverting the polarity of the signal, that's a pretty surefire way to make that happen. You may have accidentally clicked the reverse polarity button in Cubase on the S track or it disables when you disable the De-esser.

Now, an actual delay (or delay compensation) would probably NOT create such a stark problem because any offset will only affect certain frequencies (comb filtering on the L and R channel would probably be aparent)

But IIRC, L + R = M and L - R = S (we're going to use hard numbers right now but they constantly shift in the computer)

If L = 3 and R = 4, M = 7 and S = 1. If you accidentally inverted the polarity of M, you'd have M = -7 and S =1 and you probably wouldn't really hear that.

But when you apply the same formula to get the result, you have M + S = R and M - S = L

7 + 1 = 8 and 7 - 1 = 6 (divide by two for the cutting in half total volume from 2 speakers to 1, same as a -6db reduction in level) 4 and 3.

Now if you inverted the 7, let's see what sorts of wacky we get:

-7 + 1 = -6 (-3) and -7 - 1 = -8 (-4)

Because it's almost inaudible if you invert the polarity on a solo track (or in this case, the entire result) you wind up with the inverse of L on R and the inverse of R on L.

so I hope that was somewhat helpful.

Keep in mind, this would be 1 sample (snapshot of waveform at a point in time) in this example but as samples would keep coming with different values, the same inherent issues would arise:

R + L = 2M, R - L = 2S.

M + S = 2L, M - S = 2R

-M + S = -2R, -M - S = -2L

[macc]

Now that's more like it
Undersood, and thanks for that

what could cause that? Well if your De-esser is inverting the polarity of the signal, that's a pretty surefire way to make that happen.

Now this is the thing - could/would a minimum phase filter bank do that? I suspect so. My physics is too rusty to do the maths though

I've not accidentally clicked anything btw

Thanks!

[donsolo]

[QUOTE=macc;2678872]
Now this is the thing - could/would a minimum phase filter bank do that? I suspect so.
QUOTE]

Keep me posted on that!

Glad I could help and I'm sorry about nerding out on you there.

[macc]

Originally Posted by donsolo

I'm sorry about nerding out on you there.

Haha! Don't be silly - it's exactly what I was after. I kind of expected a few more geeky answers really... Call yerselves ME's?!?

[donsolo]

Originally Posted by macc

Call yerselves ME's?!?

The term is used somewhat loosely on here.

I'm just glad I get to come on here and talk to the real MEs.

[macc]

Same here - I was hoping they'd geek a bit, but they're too busy being real MEs and suggesting ways to fix the problem, rather than explaining it

[donsolo]

so was that the issue?

[macc]

My brain's still processing it... I am sure it will come up with an answer at some point

Of course I could just ask the UAD doctors. Maybe I'll do that actually