Cleanest, least colored freeware compressor? POLL ADDED!!!!

[bmanic]

.. and that test is pretty much flawed. Clean doesn't have to mean that it has less distortion than others. Clean in my humble opinion means transparent, which is directly related to how it reacts to the audio material it is processing. Not only some steady state sine tone. This has been discussed before, in the case of limiters, where Mr Kjaerhus made a pretty controversial (and completely flawed) demonstration where his limiter had the least distortion when limiting a sine tone. However, he failed to mention that he didn't tweak each plugin for their optimum performance and the test also completely failed to demonstrate how the limiter reacted to transients which is much more important than some distortion that can not be completely avoided and becomes irrelevant unless abusing the limiter/compressor.

Just my thoughts on the subject..

Cheers!
bManic

[MrM]

I can't disagree with all the comments.

Notice, I did write 'a small' test. I didn't say that it was an broad, extensive test.

And, I also wrote that for me, it counts far more how easy it is to operate a compressor.

Testing with transients is a good idea.

But: I also read somewhere (in this thread) that each compressor has its own sound, and sometimes you want that sound. Limiting yourself to only one - very clean - compressor is a bad idea in my book, expecially with all the free ones out there.

[Rottweiler]

can sy gimme a link to the vanilla comp? as ia said in an other thread i cant find it

[jens]

it seems you're out of luck...

I looked for a license-file but didn't find any - I'm not sure whether I accidently deleted it or if it came without one - the manual (which I have) doesn't touch this point - sorry.

[LOSER]

.. and that test is pretty much flawed. Clean doesn't have to mean that it has less distortion than others. Clean in my humble opinion means transparent, which is directly related to how it reacts to the audio material it is processing. Not only some steady state sine tone.

Leaving out the definition of clean (which in my opinion is no distortion/atrifacts) I can totally see myself agreeing with you. BUT how can a compressor be clean (or work any better) on a COMPLEX signal when it fails to compress a STEADY sinewave? (I hope you get my point.)
I mean the above test was sort of over kill using such short wavelengths, but I'd say for a 1k sinewave any compressor should be clean (meaning no distortion nor other artifacts) IMO.
Also I know that distortion is desirable, but I'd rather know when distortion get's applied to choose when to use it and otherwise add it seperatly myself.

Jens: Vanilla Compressor: http://www.gersic.com/plugins/index.php?daPlug=726

[Rottweiler]

it seems you're out of luck...

I looked for a license-file but didn't find any - I'm not sure whether I accidently deleted it or if it came without one - the manual (which I have) doesn't touch this point - sorry.

license file?

i thought it was FREEWARE

gialol



@LOSER: thx for the link , brother

previously i found it, but when i dled the file it wasnt a completely downloaded and i didnt find the link aagain

thx

[aciddose]

"I'd say for a 1k sinewave any compressor should be clean"

when you do this test you're measuring the quality/slope of the envelope filter. the same issues apply as always, with a perfect butter-worth q giving the sharpest edge without pass/stop band ripple. http://en.wikipedia.org/wiki/Elliptic_filter lists the common types.

the compressor with the best filter slope / stopband ripple is going to give you the best results for your 1k / other tests, however this same compressor will have terrible lag in it's envelopes meaning you'll have a high minimum attack time. there are a ton of issues where the solutions are always trade-offs. like has already been said you cant have the ideal compressor since there are too many inter-dependent factors to deal with.

[LOSER]

"I'd say for a 1k sinewave any compressor should be clean"

when you do this test you're measuring the quality/slope of the envelope filter.
[...]

Yes true, but can we just assume for 1 second that we are NOT bound to design concepts and their limitations here? (May it be modeling of analog circuits or plain simple digital dsp.)

I mean sure there will be ripple, jitter and whatever all that stuff means anyway... BUT given a PERFECT envelope follower it WILL give you a straight line for a 1kHz (also for a 1Hz sinewave) that is EXACTLY the amplitude of the signal, no? And isn't this exactly what this is about? Finding the cleanest compressor? Or am I in the wrong thread?

Anyway there are trade offs between cleanness and responsiveness. But we are after the cleanest and this is the one that yields the least distortion over the broadest spectrum (and this includes 1k ), though you could also say the least distortion in the frequency spectum of its application would suffice.

I stand by that point, though I admit that it doesn't say anything about the useability of the compressor itself.

In the end its up to everyone to choose the right tools for the right job at the right time, anyway.

[aciddose]

a perfect envelope follower is impossible - either it will respond poorly at high frequencies or poorly at low frequencies. even a follower created by peak&hold reset on each zero crossing of an analyzed fundamental will still produce "distortion". this "distortion" you're talking about is the result of modulating the signal's amplitude by some filtered version of itself.

an envelope follower can easily respond perfectly to a 1khz sine - it need only be a simple half-rectified peak&hold. however, such a design will not change state again after the sine has been removed. it will only lock on to the highest input amplitude and hold at that level forever. so, this shows the fundamental flaw in the logic you've applied to create this test. the only thing you are testing is the quality/slope (by the way, quality means 'resonance' and other things, look it up) of the envelope follower's filter, nothing else.

[jens]

it seems you're out of luck...

I looked for a license-file but didn't find any - I'm not sure whether I accidently deleted it or if it came without one - the manual (which I have) doesn't touch this point - sorry.

license file?

i thought it was FREEWARE

Freeware usually also has a license.

[bmanic]

Leaving out the definition of clean (which in my opinion is no distortion/atrifacts) I can totally see myself agreeing with you. BUT how can a compressor be clean (or work any better) on a COMPLEX signal when it fails to compress a STEADY sinewave? (I hope you get my point.)
I mean the above test was sort of over kill using such short wavelengths, but I'd say for a 1k sinewave any compressor should be clean (meaning no distortion nor other artifacts) IMO.
Also I know that distortion is desirable, but I'd rather know when distortion get's applied to choose when to use it and otherwise add it seperatly myself.

Jens: Vanilla Compressor: http://www.gersic.com/plugins/index.php?daPlug=726

Sure, distortion is a problem when it is too prominent but in my humble experience, what makes a compressor "transparent" (that is my 'clean'.. when we mention "pull up a clean compressor" in the studio, we mean transparent. We are not talking about how it distorts or doesn't distort the signal) has very little to do with distortion. Heck, there are some classic tube compressors that have plenty of distortion, yet they manage to sound much more transparent than any digital equivalent. Now how is this possible?

Simple, it has everything to do with how the attack and release reacts to the incoming program material. So, if one codes a plugin with simple linear attack and release without any program dependency you will not end up with a "clean" compressor even though it had the best distortion figures in the world. It would sound un-natural and probably pump badly at heavier distortions.

This is what makes compressors so interesting in my opinion. There are so many variables to consider, where distortion surely is a part of it, but not at all the most important one.

Then again, one might ask: What is transparent compression? It all becomes pretty subjective whereas distortion can easily be measured and thus becomes objective fact.

Cheers!
bManic

[dickiefunk]

Hi. Thanks for all the info and votes. Quite a bit of the technical info went above my head but will be handy as I learn more!

Looks like the Kjaerhus Classic Compressor is the best option. Will also check out ReaComp and Jeroen Breebart PC-2.

[brok landers]

Sure, distortion is a problem when it is too prominent but in my humble experience, what makes a compressor "transparent" (that is my 'clean'.. when we mention "pull up a clean compressor" in the studio, we mean transparent. We are not talking about how it distorts or doesn't distort the signal) has very little to do with distortion. Heck, there are some classic tube compressors that have plenty of distortion, yet they manage to sound much more transparent than any digital equivalent. Now how is this possible?

Simple, it has everything to do with how the attack and release reacts to the incoming program material. So, if one codes a plugin with simple linear attack and release without any program dependency you will not end up with a "clean" compressor even though it had the best distortion figures in the world. It would sound un-natural and probably pump badly at heavier distortions.

This is what makes compressors so interesting in my opinion. There are so many variables to consider, where distortion surely is a part of it, but not at all the most important one.

Then again, one might ask: What is transparent compression? It all becomes pretty subjective whereas distortion can easily be measured and thus becomes objective fact.

Cheers!
bManic

bmanic, you're right on all points, just lemme add something i find important right on the quote marhed red:
the reason why some tube comp's sound transparent, although there is distortion applied, is "because" there's distortion applied ... it's just that the tube is tweaked to only add _certain_ harmonics, that please the sound ... in good tube compressors (or any other good distortion-applying compressors) the key is, that the ear doesn't detect the distortion as "audible distortion", but as an integrative part of the signal ...
the problem in digital domain is, that this distortion would apply heavy aliasing, if not oversampled in a hughe ammount ...
bandlimiting (via fft>ifft) can't be a solution, as this would add an uncomftable delay to the processing, and although that could be compensated by the host, there's still the cpu usage, as the bandlimiting has to be realtime ...
that's why digital models of analog compressors mostly, if they add _wanted_ distortion, sound not that "creamy" ... my english sucks, but you get the point ...

also, as a sidenote, often cheap compressor plugins tend to distort because of the detector of the envelope follower isn't updated every sample, so if the times are too short, there's jumps/steps between the values the envelopes runs through ...
this can't happen in the analog domain ...

[jens]

Will also check out ReaComp and Jeroen Breebart PC-2.

make sure you switch to the complex (=psychoacoustic rocket-science) mode when you check out PC-2 (there's a little button left from the gain-fader) - by default PC-2 is set to the conventional-compression mode...

[LOSER]

a perfect envelope follower is impossible - either it will respond poorly at high frequencies or poorly at low frequencies. even a follower created by peak&hold reset on each zero crossing of an analyzed fundamental will still produce "distortion". this "distortion" you're talking about is the result of modulating the signal's amplitude by some filtered version of itself.

an envelope follower can easily respond perfectly to a 1khz sine - it need only be a simple half-rectified peak&hold. however, such a design will not change state again after the sine has been removed. it will only lock on to the highest input amplitude and hold at that level forever. so, this shows the fundamental flaw in the logic you've applied to create this test. the only thing you are testing is the quality/slope (by the way, quality means 'resonance' and other things, look it up) of the envelope follower's filter, nothing else.

Yes true, but can we just pretend for 1 second that we are NOT bound to design concepts and their limitations? Because if you give me a signal plot I can totally draw in the perfect envelope. So just because it is not realizable via simple filters doesn't mean it is impossible.

So I think that my point is still valid that when assumed that cleanest, and least color means doesn't add distortion nor modulation or other artifacts, that then but only then the winner will be the - and may I quote myself - "one that yields the least distortion over the broadest spectrum[...], though [one] could also say the least distortion in the frequency spectum of its application would suffice".
Also regarding your modulation theory thing ... IMO should there be no modulation going on on a steady signal since its envelope is constant, and if it still occurs, then the filter isn't good, that means it has this thing, how was it called again, wait let me look it up, not steep enough roll off or a too high frequency.

Next the fundamental flaw is in your logic, how about you try not to depict me as a total moron, because noone is every gonna use a infinit long peak hold. How about this, using 3 peak holds in parallel that reset in a round-robin fashion, like proposed by Harry Bissell and Royal Oak in their 2002 by EDN published article "Envelope follower combines fast response, low ripple"?

But someone that had a talk to you some years ago told me that a discussion with you is lost and a waste of time anyway, so I will remain silent and distortion in compressors is good, whatever.

But this was never about the cleanest (as in lowest distortion) compressor anyway, but rather the typical "which compressor sounds most Xish", right?

BTW bmanic this could be due to the distorion that hardware adds, which are mostly harmonics, which sound more pleasing to the ear than inharmonics.

P.S. This kinda reminds me of when everyone including Sascha Eversmeier told me my Zero Crossing Limiter wouldn't work and suck and distort, though still it got more votes than the W1 in a blind limiter shootout. Granted only 11 people paricipated. But well ... I was trying to get measurements of distortion on attack and relase phases of compressors, though that might get me flamed, so I'll just use it for my own research.