ABX Compressor Test VST

[LOSER]

In order to do some research on dynamic range compression, mainly about what sounds better, I build a little VST compressor plugin that has 18 different modes that switch between different compressor implementations (mainly different attack and release curves). The idea is to basically have as many people evaluate the different modes and see if there is a consensus for what works best for what kind of material ... or something a long those lines.

Anyway I can't ofter anyone anything for their efforts of ABX testing those different implementations except for - well the VST ... Come on! That's 18 different compressors in ONE! So have at it.

http://audio.michael-gruhn.com/files/AB ... orTest.zip

All modes implement a feedforward, peak compressor that has no additional filtering. The txt in the zip also has some more details.

I basically had in mind that people could please comment on what modes they like best and on what material. General and negative comments (about the sound of the different modes) are also welcome.

Appreciate your attention and happy ABX'ing or not as I'm not gonna force anyone to do it .

[airon]

Have you ever thought of venturing in to new territory, like this hardware unit here: http://www.charteroakacoustics.com/prod ... etails.htm

You've got the experience to maybe pull this off.

What's one more in a crowd of 18 . Testing now.

[Debutante]

awesome learning tool to be had. must try. THANKS!!!!!!!

[Debutante]

f**king Awesome!

[LOSER]

f**king Awesome!

What mode exactly is "f**king Awesome!"?

[BFunKu]

This has got to be the Most versatile Compressor I have ever seen! 18 different modes! I don't think there is another compressor with that many modes?

[funkadil]

Haven't tested it yet. But is it entirely a mystery what the modes actually are, or does it tell you?

[michu]

Haven't tested it yet. But is it entirely a mystery what the modes actually are, or does it tell you?

http://en.wikipedia.org/wiki/ABX_test

i think the whole idea is that you don't know which mode is which

[funkadil]

ahhh, I missed the significance of ABX in the OP.

[Compyfox]

If it's only attack/release in the different modes, what benefits does it except for different "sound", like the LA-2A, where you can't setup attack/release either? According to KVR's spec, it's just a FF-Comp with Sidechain.


No disrespect or anything (definitely love the transient desinger, however the old, non MS one).

[LOSER]

If it's only attack/release in the different modes, what benefits does it except for different "sound", like the LA-2A, where you can't setup attack/release either?

Sorry if I maybe wasn't as clear as I should be but this is a _research_ project more than anything (though it features some full functional compressors and then some). So to prevent further confusion here is a quote from the readme.txt

ABX Compressor Test is a research VST plugin that implements 18 different
compressor modes, that is 18 more or less significant different algorithms. It
was created to research how a good compressor should sound (and therefor how it
must be designed).

All compressor designes are feedforward and peak detecting. The sidechain
is always not filtered. The signal path is also unfiltered, that is it contains
no sound enhancing of any kind (except for the gain adjustment performed by
compression).


Different modes:
----------------
The different modes comprise different envelope follower types, different
transfer functions as well as different interopreation between the internal
parts of the compressor. The modes are numbered from 1 to 18, the literals in
parenthesis hint at the difference and can be interpreted as follows:
HQR = High quality transfer function (ratio)
LQR = Low quality tranfer function (ratio)
S = Smooth operation (release)

My initial intend was to get an _unbiased_ evaluation of various compression designs. What I found in private testing (same approach) however was that the more is revealed about the different modes (that is designs) the more biased and tainted the test becomes, as statements like mode 3 is virtually identical to ReaComp with classic attack and 0 RMS size, etc ... inevitable lead to bias, placebo effect, etc ... all of which isn't desired.

So to sum it up the VST is a compressor with threshold, ratio, attack, release and output gain, and the "mode" parameter merely switches between different designs, which of course have different sounds, use different amounts of CPU time, etc ... so you can see it like one of those "which compressor sounds best" threads but only this time there is no bias, since everyone has the same black box.

Feedback was my main criteria for making this so public, because private testing in a small group doesn't give representable results.
Maybe I should have posted it in the "DSP Development" section.

Anyway maybe it was a big mistake to start this "Frankenstein" experiment.

[Compyfox]

I don't think that this was a mistake. Different people have different opinions and needs. I was just wondering.

So the circuity switches are only "internal" and you still have full control over the compressor itself. Much like Blockfish back in the day, where you could change all kinds of parameters, only that this one is "hidden" this time.

Sure might be an interesting concept, and I'm sure a lot of people will test that. However I count myself to the lazy fraction of the users - I want documented stuff rather than "blind-testing".

But who knows, maybe we all are fooled in the game and this whole thing is placebo in terms of "soundshaping". Then again... this is the fun of the game... we don't know it.



BTW:
Any chance to see an update on your old TransientDesigner (the non M/S version)? Nothing special, just rendering in VST2.4+ Standard maybe. I remember it was VST2.3 - still it's my most go-to tool for transient shaping (don't like anything else to be honest). Never really liked the M/S version though, too many parameters and not the same response. At least on my end.

[DuX]

I wouldn't say it was a mistake . An excellent idea, actually!

Cheers!

[LOSER]

So the circuity switches are only "internal" and you still have full control over the compressor itself.

Yes.

Much like Blockfish back in the day, where you could change all kinds of parameters, only that this one is "hidden" this time.

I'd say no, if in anyway however it is comparable with Blockfish's VCA/Opt switch because it switches the design of the compressor (or parts like the transfer function) while the extra controls (I think you are referring to those) of Blockfish let you adjust the compressor behavior itself and not the compressor, or am I mistaken.

Sure might be an interesting concept, and I'm sure a lot of people will test that. However I count myself to the lazy fraction of the users - I want documented stuff rather than "blind-testing".

Like said before once it gets documented people get biased apart from that there isn't really a way to document it because the different modes are actually different compressor implementations and I wasn't joking in my previous post that mode 3 is mostly very identical to ReaComp with the mentioned settings. So the only way to really document - so that people can make sense with the documentation (that is a non techy doc) - is to tell to what other compressors the different modes are similarly to, which would defeat the entire purpose of this test ... anyway seems to be defeated already .

BTW:
Any chance to see an update on your old TransientDesigner (the non M/S version)? Nothing special, just rendering in VST2.4+ Standard maybe. I remember it was VST2.3 - still it's my most go-to tool for transient shaping (don't like anything else to be honest). Never really liked the M/S version though, too many parameters and not the same response. At least on my end.

Hope this is the correct one (since I made so many of these): http://loser.asseca.com/downloads/Trans ... -09-01.zip

EDIT: Retrospectively I think I should maybe have labeled the "mode" parameter "Compressor".

[Compyfox]

I'd say no, if in anyway however it is comparable with Blockfish's VCA/Opt switch because it switches the design of the compressor (or parts like the transfer function) while the extra controls (I think you are referring to those) of Blockfish let you adjust the compressor behavior itself and not the compressor, or am I mistaken.

So it's like the URS Channel Strip Pro too, where you can also select different kinds of compressor "characteristics", but the attack/release is untouched...

Like said before once it gets documented people get biased apart from that there isn't really a way to document it because the different modes are actually different compressor implementations and I wasn't joking in my previous post that mode 3 is mostly very identical to ReaComp with the mentioned settings. So the only way to really document - so that people can make sense with the documentation (that is a non techy doc) - is to tell to what other compressors the different modes are similarly to, which would defeat the entire purpose of this test ... anyway seems to be defeated already .

Don't think that it is defeated - because every person thinks different on EACH compressor (trust me, even about EQs, even if they built up similar). I however learned my lesson with Crysonic and that undocumented mojo. Nothing against you, really. But I'm sure that compressor IS something for people who like to have more than one compressor in one - even if it's only FFC.

So again, cudos for that.

Hope this is the correct one (since I made so many of these):

I need to test that. If I remember, one was called TranShaper, one TransientShaper. One was the core version (attack and release settings only) and the other one was the M/S version. Dunno however which one was which. But I always revert to the "standard" version if I need a transient designer.

EDIT: Retrospectively I think I should maybe have labeled the "mode" parameter "Compressor".

Why not name it "compressor mode"?