Different types of compression, SSL vs 1176 vs 1a2a vs DBX160 ect...

[manducator]

Yes. What of?

I'm happy with anything you offer...

[X2]

stillwell audio bombardier can model any compressor I know of and the bomb mode sounds exactly like the softube FET all buttons sound when the attack is set to 0 and the drive is alway up, sag controls the depth and length of the release stage,especially in feedback mode and the 5 selection EQ SC is Genius. Also have to admit that Density mk3 is as good as any Vintage Compressor emulation going and the flexibility of its controls is quick and easy to get any sound your after.

[meloco_go]

A logging amp is a waveshaper which generates y=log(abs(x)). You use it to convert a signal (which is a voltage) to a dB measurement.

So they are used to provide user with some meaningful reference to a threashold level, gain reduction going on etc?
Are they also used on compressors with controls which are not calibrated in dBs?

I'd love to know if you find the same thing in cans, but further, if you find it with renders, we can delay compensate the renders, and phase cancel them!

You can find quite a lot of them over the internet.

[DaveGamble]

A logging amp is a waveshaper which generates y=log(abs(x)). You use it to convert a signal (which is a voltage) to a dB measurement.

So they are used to provide user with some meaningful reference to a threashold level, gain reduction going on etc?
Are they also used on compressors with controls which are not calibrated in dBs?

No, they're used to calculate the GR signal.
In a basic DBX style feed forward circuit, you take the input signal, log it, subtract the threshold, half-wave rectify, and now gain adjust by ratio. This is a series of circuit steps implementing a gain reduction equation. Next comes attack&release, and then you have the finished GR signal which will drive the VCA.

They're implicitly used by any unit that has log attack/release shape. You might still use one in something with linear curves, for instance if you're doing AR first, then into a logging amp, etc. I've seen that in some really esoteric gear, and it's a lot of fun. Precision with that old school linear vibe.

I'd love to know if you find the same thing in cans, but further, if you find it with renders, we can delay compensate the renders, and phase cancel them!

You can find quite a lot of them over the internet.

Searching for such things is not good fun for me. You get halfway in, and discover that someone left an EQ on one of them, or did something frightfully weird. I'm entirely happy to help you analyse, but I'm not going looking, sorry. And that's in no small part because I genuinely don't believe there's anything to find.

You will credit that I'm someone who gets sent examples of things, yes? And I did design a rather comprehensive dynamics processor. And I'm VERY much aware of the "software compressors and lower mid" meme. Yet, I've never head it, nor received a complaint of it with compassion, nor been sent ANY examples of it. Literally, no one has ever called me over to a thread to explain it.
Obviously, I'm a scientist, so should evidence appear, I'll respond to it, but as of this time, 12years into designing software compressors (and analysing/modelling hardware), this claim appears to be a foundless meme, perhaps propagated by some hardware designers, which has stayed around, despite a complete lack of empirical evidence.

Dave.

[DaveGamble]

As caveat to the above, obviously I can't rule out that there actually IS a plugin compressor out there which is somehow broken, and depleting lower mid, but I don't know what it is.

For the claim that "all software compression" vs "all hardware compression", I don't think you could sustain it.

As further dialogue, I am aware of a brief movement in software (which has now been almost entirely eclipsed, thankfully), where Udo Zolzer published a DAFX book containing a "design" for a software compressor which performed extremely poorly. This design appeared to be adopted by some developers for a while, there was a corresponding "software compressors sound bad" reaction, and then the design was dropped. Basically there was an AR stage which created 1st order discontinuity in the GR signal, and it sucked. But that was more than just influence over the lower mid. It made compressors that sounded bad, period.

Dave.

[lacandon]

the dbx 160 vu a is one trick pony vca comp mostly for snare kicks and sometime bass lines . Vary mu comps are coloured comps mushy sounding used mostly for vibe and wider overall sound. The Ni Vary Mu is so far away from the true i do not suggest it ,alex b fenix comp is a cool vary mu comp u can get the vibe from it .The SSL G comp is industry standart bus comp used mostly to glue the sounds , my choice for this type comp will be Sknote Mixbuss 4000 very solid or u can get even terrific bus comp from Cupwise and their new Neve bus comp emulation witch is most close to analog gear i ever heard.
This Video can be very interesting for you to hear what most of them doing to the source p://

[meloco_go]

No, they're used to calculate the GR signal.
In a basic DBX style feed forward circuit, you take the input signal, log it, subtract the threshold, half-wave rectify, and now gain adjust by ratio. This is a series of circuit steps implementing a gain reduction equation. Next comes attack&release, and then you have the finished GR signal which will drive the VCA.

Ah, I don't read schematics very well=) I thought the compression curve was a non-linear element, and then some component to get a difference signal between it and original.

They're implicitly used by any unit that has log attack/release shape. You might still use one in something with linear curves, for instance if you're doing AR first, then into a logging amp, etc. I've seen that in some really esoteric gear, and it's a lot of fun. Precision with that old school linear vibe.

I thought log attack/release comes from filtering key signal? Or this is what you mean? How attack/release worked on old gear? I imagine opto's have attack/release shapes "built-in" in the cell properties. What about others?

Searching for such things is not good fun for me. You get halfway in, and discover that someone left an EQ on one of them, or did something frightfully weird. I'm entirely happy to help you analyse, but I'm not going looking, sorry. And that's in no small part because I genuinely don't believe there's anything to find.

Yeah, fair point=)

You will credit that I'm someone who gets sent examples of things, yes? And I did design a rather comprehensive dynamics processor. And I'm VERY much aware of the "software compressors and lower mid" meme. Yet, I've never head it, nor received a complaint of it with compassion, nor been sent ANY examples of it. Literally, no one has ever called me over to a thread to explain it.
Obviously, I'm a scientist, so should evidence appear, I'll respond to it, but as of this time, 12years into designing software compressors (and analysing/modelling hardware), this claim appears to be a foundless meme, perhaps propagated by some hardware designers, which has stayed around, despite a complete lack of empirical evidence.

Well, but there's also components like transformers or amps, which are not part of GR circuit but color the signal. Maybe that is what people are referring?

[APZX]

DaveGamble,

I am not one to go out of my and understanding the exact inner workings of a compressor . . ., yet (starting to cut my teeth into circuit design because a big future goal that probably won't happen lol). Anyway, I do find I select compressors based more on the way the compressor behaves to material and that brings up a question that I have yet to figure out. It deals with a Vari-Mu style compressor, the classic Gates Sta Level. Now, my question is not so much about the way it compresses, which if memory serves it is a pretty simple program dependent circuit which has an interesting attack characteristic (something like a cross between an 1176 and a VCA depending on what you throw at it) but otherwise behaves more like an Opto. The question I have is why approach compression that way (I get that it was more meant as a limiter and it was like '56)? What advantage does that offer in terms of circuit design or even compression characteristics over say a transistor based circuit?

[softska]

this thread makes my coffee taste that much better.

I swear Dave, if Compassion shown up in my life years earlier it would have convinced me to switched over to EE major

[DaveGamble]

No, they're used to calculate the GR signal.
In a basic DBX style feed forward circuit, you take the input signal, log it, subtract the threshold, half-wave rectify, and now gain adjust by ratio. This is a series of circuit steps implementing a gain reduction equation. Next comes attack&release, and then you have the finished GR signal which will drive the VCA.

Ah, I don't read schematics very well=) I thought the compression curve was a non-linear element, and then some component to get a difference signal between it and original.

Yes, it is, but it's a very specific, controllable nonlinear element!
Now, mind that here I'm just talking about SOME designs; you can do this however you like, depending on how "perfect" you want your compressor to be, but essentially, the nonlinear element for the "compression curve" (which I am interpreting as threshold and ratio circuitry) is y=ratio_coeff*max(0,20*log10(abs(x))-thresh_db);

This IS a static nonlinearity, but to do it accurately, you do it in stages (work from the inside out): rectify, log, subtract, half-wave rectify, gain stage.

Now, to get soft knees and such, you CAN add a waveshaper (try an uncompensated half-wave rectifier!) in at pretty much and stage in this process. You can do it later too, I.e. after the AR stage, and right before the VCA control input.

You can do pretty much whatever you like and get SOME kind of result, which is why there are so many different sounding compressors.

They're implicitly used by any unit that has log attack/release shape. You might still use one in something with linear curves, for instance if you're doing AR first, then into a logging amp, etc. I've seen that in some really esoteric gear, and it's a lot of fun. Precision with that old school linear vibe.

I thought log attack/release comes from filtering key signal?

Ah, didn't really explain this bit. It's not straightforward.

"Log" means that attack/release curves show a 1st order exponential type behaviour in attenuation in decibels.
"Linear" means that attack/release curves show a 1st order exponential type behaviour in voltage.

It's named ass-backwards; it's the type of curve you get when your AR stage is working in that domain.

Or this is what you mean? How attack/release worked on old gear?
I imagine opto's have attack/release shapes "built-in" in the cell properties.

Yes, an incredibly fast one that gets superposed over the AR stage.

You will credit that I'm someone who gets sent examples of things, yes? And I did design a rather comprehensive dynamics processor. And I'm VERY much aware of the "software compressors and lower mid" meme. Yet, I've never head it, nor received a complaint of it with compassion, nor been sent ANY examples of it. Literally, no one has ever called me over to a thread to explain it.
Obviously, I'm a scientist, so should evidence appear, I'll respond to it, but as of this time, 12years into designing software compressors (and analysing/modelling hardware), this claim appears to be a foundless meme, perhaps propagated by some hardware designers, which has stayed around, despite a complete lack of empirical evidence.

Well, but there's also components like transformers or amps, which are not part of GR circuit but color the signal. Maybe that is what people are referring?

Sure, but then you'd get the same thing while the compressor was doing nothing to the signal.
Things are NOT that interactive, or they wouldn't work. Input stage, compressor, output stage.

Dave.

[SWAN808]

If you can demonstrate with some examples, we can (and should!) evaluate this, but I suspect the reality is that when you're monitoring software, your head is in the sweetspot, while when you're monitoring the outboard, your head is outside the sweetspot, and you're adjusting timings accordingly. I support this claim with the observation that half a wavelength of low mid is going to be in the order of a foot or two, which is probably how far you have to move your head to look at a unit you're adjusting.

Dave.

Hi Dave thanks for the answers!

In terms of this difference in sound between the low mid punch Im pretty sure Ive heard it with almost every software compressor Ive tried / heard...although its quite subtle it affects the sense of punch and impact. Im going to have to dig out some samples if I can find them to help explain...but I remembered I asked sascha a similar question a while back and he gave me an interesting answer...although I dont think I fully understood it!

http://www.kvraudio.com/forum/viewtopic ... compressor

ok here is one example with dbx compressors...its close but to me it feels like there is more convincing and solid low mid punch in the hardware...

http://www.gearslutz.com/board/8048362-post23.html

[deastman]

Dave,

I was wondering if you could comment on the venerable Alesis 3630. I know its basically garbage, but I've been using them in non-music A/V applications for a long time, and know it better than any other compressor.

[SWAN808]

Hi Dave these are probably better examples:

SoundonSound media from their Portico plugin review tested against the hardware. The plugin seems excellent and as documented in the review they spent a good deal of time tweaking them...only when I directly switch between the loops whilst playing do I hear the slightly differing quality. To my ears its hard to tell sometimes the high end is more open, but on the 3 examples Ive included - all of them I was able to distinguish slightly more low mid punch and generally a sound that was a fraction firmer/together in the HW example...when listening Ive just been using my HD650 Headphones today but they are decent for finding detail possibly more so than a so-so monitoring set-up...

https://dl.dropboxusercontent.com/u/799 ... amples.zip

[FeelingMachine]

(refering to dbx160)

That was a nice comparison. I at first heard a more solid punch to the hardware version, however, i did a rigorous blind test and other than that inital kick drum, i could not distinguish the two.
I felt the inital kick of the hardware loop hit harder (which is probably just the silence to impact effect) than the software which would lead you to believe the rest hits as hard. But by strictly listening to the rest of the loop i couldn't hear a difference.
That for me is enough to use the software. Im using Sennheiser 650s and im not really concerned for any detail more subtle than that. The whack and punch was there - for me thats good enough for a pro mix.
I must say, i still use and believe in the quality of hardware but there clearly is software that is more than capable.
I also think that Dave has demonstrated to us that a DBX comp such as the 160 is pretty straight-forward in design and this may explain why it has been emulated so well above other designs.

@Dave, i think the dsp world is fortunate to have the likes of you, Sean Costello, etc around !!

I now also truly understand COMpassion lol ! I knew it was flexible, but listening to you i see that it really is a bunch of components that allow us to build any compressor design we like.

Which leads me on to one question that im not sure has been asked in this thread ; saturation, harmonics, distortion, etc from compression. Basically, what is the 'story' here. Does it exist in compressors, does COMPassion do it or how would i incorporate it. I know its a long one so if you have a link im more than happy to read.

+1 on hardware presets for COMpassion !

[SWAN808]

I agree the dbx models are close and pretty good...in fact in an ABX test machine its really quite tough to distinguish...but once you get your ear in - the kick is one thing, but also the way the compressor clamps around the clap...in both examples its just slightly more enveloping the sound and holding a bit more bass...really if you are good at mixing its not going to cause a huge change, but still its a characteristic I tend to hear...

Im starting to wonder if the fact that software carries all freq esp subs but hardware filters out some sub and higher freq making it sound punchy...(as well as a bit of saturation?)...