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

[MrM]

OK, a small test.

I'll refer to the compressors as numbers:
#1 = BuzComp
#2 = Compressive (Lite)
#3 = Kjaerhus Classic Compressor
#4 = ReaComp
#5 = Vanilla Compressor:
#6 = MjCompressor
#7 = JB PC-2

Test 1,
20 Hz going into the comp, settings:
Treshold @ -20 db, ratio @ 1:6.7 no knee, attack @ 20 ms, release @ 200 ms, auto make up gain.

#1 - I am afraid to name this: distortion @ -21 dB !!!
#2 - distortion @ -50 dB
#3 - distortion @ -47 dB (with hard knee or -37 dB with soft knee)
#4 - distortion @ -45 dB
#5 - no distortion
#6 - distortion @ -50 dB
#7 - distortion @ -46 dB

#5 is best in this test.

Note about #5:
The build in limiter of this one is crap, better not turn that one on.
Personally, I have trouble operating some buttons; the max make up and detection knobs have only two postions but are hard to engage.
This is because the host I use, but the problem would go away if there was a GUI.

Test 2,
40 Hz and 185 Hz together, same settings (I am lazy!).

#1 - distortion @ -26 dB
#2 - distortion @ -54 dB
#3 - distortion @ -54 dB (avoid the soft knee: -39 dB)
#4 - distortion @ -48 dB
#5 - no distortion (actually there probably is some, but I am having problems operating the knobs, so I can't really test this in a valid way, I am therefor disqualifying the test results for this plugin, but so far it is a clear and transparant VST)
#6 - distortion @ -47 dB
#7 - distortion @ -50 dB

#5 is best in this test, #4 is runner up.

Test 3,
20 Hz and 50 Hz together, same settings (I am lazy!).

#1 - distortion @ -20 dB, ouch! Or, yeah, if you want some color
#2 - distortion @ -32 dB
#3 - distortion @ -27 dB
#4 - distortion @ -31 dB
#5 - distortion @ -40 dB
#6 - distortion @ -32 dB
#7 - distortion @ -33 dB

Best, as in most transparant: Vanilla

Would I use it? Hell no!
Classic Compressor is nice to handle, it is my personal favorite
Why, because it is better to operate. Transparant is nice, but if it is hard to operate a VST, the VST is still no good for me

[jens]

Compressive from Martin Eastwood.

http://martineastwood.com/?p=3

clean?? imo that thing sounds really analog coloured, and that in a very high quality ... one of my go-to's when it comes to compressors ...

exactly - this thread is full of bull - the most transparent compressor (which applies some kind of - really working - rocket-science) hasn't even been mentioned yet...

http://www.jeroenbreebaart.com/audio_vst.htm

[dickiefunk]

Thanks for the test Mr M. Could you test the Jeroen Breebart PC-2 compressor and Barricade?

Thanks

http://www.jeroenbreebaart.com/audio_vst.htm

[LOSER]

OK, a small test.
[...]

Nice test, though may I suggest some things.
Test with a 1kHz sine wave too, because right now you test only the lowend and most audio effects are aimed to work on the whole frequency content and not just the lowend.

Well, anyway this will tell you that e.g. #4 has still reasonable distortion at 1kHz. Another thing would be sine sweeps to see how they act on the whole spectrum, e.g. here the Cockos ReaComp (default settings, with RMS size set to 0 and threshold so it yields a 6dB GR, but shouldn't matter since it is constant signal level anyway and thus constant GR etc..):



Here you can see the distortion reaches way up to the 20k range, while the Kjaerhus Classic Compressor has distortion/artifacts only in the low end.




So you need to consider what you define as transparent least colored, because as far as I can tell does you're test only ensure least coloration in the lowend, though if an effect is clean in the lowend it should be clean all over the spectrum anyway.

P.S. Before anyone asks the sinesweep picture stuff is explained here: http://www.cockos.com/forum/showthread.php?t=18471
(If anyone wants to add to the sine sweep collection, you're welcome, so when it is done I might trim all the pics and publish them on one site with other test data or if someone else wants to make a site like that take them or whatever.)

[MrM]

Thanks for the test Mr M. Could you test the Jeroen Breebart PC-2 compressor and Barricade?

Thanks

http://www.jeroenbreebaart.com/audio_vst.htm

Just added, see the posting above.

[MrM]

OK, a small test.
[...]

Nice test, though may I suggest some things.
Test with a 1kHz sine wave too, because right now you test only the lowend and most audio effects are aimed to work on the whole frequency content and not just the lowend.

...

So you need to consider what you define as transparent least colored, because as far as I can tell does you're test only ensure least coloration in the lowend, though if an effect is clean in the lowend it should be clean all over the spectrum anyway.

You are right, and I just tried to test with 1 kHz, but my internet PC lacks enough meters to test it. My DAW has plenty of meters, but I don't like to test on DAW - switch to internet PC - test on DAW - switch to internet PC to report, cause every switch requires a reboot: I run a dual boot system.

Maybe someone else will duplicate my test, and add the 1 kHz.

[dickiefunk]

OK, a small test.
[...]

Nice test, though may I suggest some things.
Test with a 1kHz sine wave too, because right now you test only the lowend and most audio effects are aimed to work on the whole frequency content and not just the lowend.

...

So you need to consider what you define as transparent least colored, because as far as I can tell does you're test only ensure least coloration in the lowend, though if an effect is clean in the lowend it should be clean all over the spectrum anyway.

You are right, and I just tried to test with 1 kHz, but my internet PC lacks enough meters to test it. My DAW has plenty of meters, but I don't like to test on DAW - switch to internet PC - test on DAW - switch to internet PC to report, cause every switch requires a reboot: I run a dual boot system.

Maybe someone else will duplicate my test, and add the 1 kHz.

Yeah that would be great!

[aciddose]

a standard distortion test is 1khz, this quantifies THD, then the average of an a-weighted spectrum. pass white noise through an a-weighting filter and compress that, test the difference between the spectra, not the difference between the transient data. in other words, take the average absolute difference between the source and processed spectra. this gives you spectral distortion.

in order to be "transparent", both should be very low. phase distortion need not be taken into account here as we're dealing with compressors. the ideal compressor should adjust phases to maximize rms amplitude to to peak amplitude ratio while providing zero harmonic / spectral distortion.

[bduffy]

http://waves.com/content.aspx?id=237&l=4

Freeware, pal; freeware. That is clean as hell, though.

mjcompressor: forgot about that. Very clean. I put Vanilla Compressor; I don't think you can detect any "color" from that.

EDIT: Ah, I see Vanilla has lived up to its name. Good stuff.

Jeez. You look at those BlueCat pics in the cokos thread, and it's clearly one of the cleanest!

[jens]

Nice test

is it? I'm not sure - but if you say so - I mean: you really know your stuff, so...


I'm just wondering what release mode Mr.M used on Vanilla? Or doesn't it matter at all for his test? And if it doesn't matter, of what use is the test?

Switching the release modes totally alters the character of Vanilla compressor, so it's not possible that they are all 'least coloured'.

Distortion surely also plays a role, but the character of a compressor is mainly in its release behaviour for me - breathing and pumping, or the lack thereof, etc.


So how valid is the test when it comes to measuring the 'colour' of compressors?

[jens]

mjcompressor: forget about that.

Will do!

[MrM]

I'm just wondering what release mode Mr.M used on Vanilla? Or doesn't it matter at all for his test?

Mode used in test: ARC-1

Does it matter to choose another mode?
With 'pump' the distortion is @ -50 dB for test 1, other modes are no problem.

[djeroen]

OK, a small test.

Test 1,
20 Hz going into the comp, settings:
Treshold @ -20 db, ratio @ 1:6.7 no knee, attack @ 20 ms, release @ 200 ms, auto make up gain.

Interesting experiment, but...

This can be quite risky. Thresholds may be defined differently
in various compressors. It would be safer to adjust the threshold
such that a fixed amount of compression is obtained.

Also attack and release times are not always
specified in the same way; sometimes they refer to a -20dB release point,
or sometimes to RT constants. The amount of distortion of a compressor is directly linked to the attack and release time constants; faster response will give higher distortion.

Then last but not least; one would have to take undesirable pumping
or distorted temporal envelopes into account to see the tradeoff between
temporal behavior and distortion...

Just my 2 cents...

[jens]

Then last but not least; one would have to take undesirable pumping
or distorted temporal envelopes into account to see the tradeoff between
temporal behavior and distortion...

that was sort of what I was trying to say...

[aciddose]

i'm not certain what the point of looking for a general-purpose "cleanest" compressor is anyway. different compressors are going to be better suited to different tasks. you cant really apply one compressor and expect it to sound good everywhere. certain times distortion is the main thing you're looking for when compressing, other times you want insane pumping.

for medical type gain control it might be better to manipulate things manually. the most accurate tool for compression might be one which takes a recorded clip and finds points where the dynamics are of interest. curved envelopes could then be drawn between these points to normalize the amplitude of the clip. no real-time processor is going to ever be able to do such a thing. i'm not sure if such a tool exists. you could always find an audio editor which allows spline amplitude envelopes and do this by hand.