compressor release times

[Satanic Smurf]

The function of the release time on compressors continues to baffle me. In all the articles that I've read I get descriptions like: Deturmines how long the compressor continues to act after the signal has gone back under the threshold. I've also read that by increasing the release time it can increase the amount of sustain.

 

So what does is meant by 'the compressor continues to act'?

 

For example if I had a compressor with a ratio of 3:1 and a threshold of -10dB, an attack time of 10ms and a release time of 50ms.

 

A signal comes into the compressor and goes above -10dB, 10ms later the compressor kicks in and starts reducing the signal as stipulated by the ratio. So a signal of -7dB becomes -9dB. Later the signal drops below the threshold of -10dB, what now? Is the signal still being reduced by the ratio of 3:1 for the next 50ms? If so, wouldn't this make the signal even more quiet How could that possibly improve sustain?

 

I understand that by reducing the transient peaks we can boost the signal without introducing clipping distortion thus making the quieter parts of the signal louder. This is how a compressor adds sustain. What does this have to do with the release time?

 

Also what happens if I recive anouther transient during the 50ms release time? I assume that the compressor just doesn't disengage

[Kiwiburger]

Its helpful to think of a compressor as something that 'compresses' audio - in other words, when working, it makes the sound lower in volume.

 

I know compressors are usually used to make things louder - but in real terms, it isn't the compression that makes it louder. It's the 'make up gain' that follows that does that.

 

A compressor does little more than what you can do with a volume control. It just makes it a lot faster and easier.

 

Imagine a volume control that is normaly full on. You are going to 'compress' the sound with this volume control. The only option you have is to lower the volume (remember the compression stage isn't the same as the make up gain stage).

 

Imagine a snare drum sound arrives. You detect that it is louder than you want (above the Threshold). You turn the volume down so it is exactly half (a reduction of 6dB - representing a compression ratio of 2:1.

 

The speed at which you reacted to this would be fairly slow - so a large part of the stick sound of the snare got through before you could react to this. A compressor can react faster than your hand on the fader, so you could dial in the amount of 'stick sound' with a compressor Attack. Fast attack will kill the stick sound by lowering it's volume straight away. Slower attack settings will let more of the stick sound through - or if it's too slow, the whole snare sound will come through unnaffected.

 

The Release is the opposite of attack. Once you realise the sound is as loud as you actually wanted, and doesn't require further compression (aka under the Threshold), you have to raise the volume back up to where you started (full gain).

 

You will probably be too slow, and maybe not return the fader to the top before the next snare hit comes through. That would be self defeating.

 

If you were fast enough, you return the fader to the top before the next hit, but in the meantime, the sound below the threshold would still be compressed. That probably isn't what you want either.

 

You probably want the compression Release to be fast enough so that as soon as the sound is below the desired Threshold, it doesn't get compressed. That requires fast Release - but not too fast. Too fast can sound clipped.

 

This is probably more obvious on longer sounds, like a cymbal crash. Ever heard that sucking sound when a crash gets smashed with a compressor, and then slowly you hear it getting louder again as the compressor slowly releases? Some people like to avoid that sound. Other people like to adjust the Release so it times with the flow of the groove.

[Satanic Smurf]

Thanks Kiwiburger, I've read your post a few times over and I think the penny in beginning to drop. This is the second time you've helped me out so, cheers I appreciate it

 

But while we are getting into the technicalities of compressors, I have more questions:

 

Does the compressor reduce the signal above the threshold or the entire signal? Example: Ratio = 2:1 with the threshold set to 10dB. A signal enters the compressor that is 12dB. Is the entire signal, the 12dB reduced or is it just the 2dB above the threshold that is reduced?

 

At what stage is the make-up gain applied? What I mean my this is: does the threshold take into account to the signal + the make-up gain?

i.e (Gain x Signal + Signal)/compresion factor

 

OR

 

is it just the signal itself that is affected and the gain is added after?

i.e Gain x Signal + Signal/compresion factor

 

My formulas could be way off (and probably serve only to confuse). I'm assuming that the make up gain is just a scalar multiplier of the signal.

 

Thanks.

[greendoor]

Makeup gain follows the compression stage, and it's just a simple volume boost (multiplier in digital terms).

 

Unless the Attack is very fast, a compressor doesn't remove transients - they slip through. And if it's fast enough to remove transients, it probably damages the sound. I think Saturation effects are better at knocking the transients off (tube/transformers/transistors/tape or models thereof).

 

This is why compressors are often used to make a sound more punchy: the attack appears greater, because the sound following it is lowered. When you add the makeup gain, the overall effect is to boost the attack portion.

 

If you are familiar with ADSR on a synth, the effect of A and R with a compressor is similar but inverted.

 

The Threshold can be considered the trigger for the Attack and Release. Both of these have adjustable time delays - imagine your fader moving at different speeds down and then back up again.

 

So although a sound may have dropped below the threshold - during the time the Release is acting, the sound will be lowered until the release period has finished. So the Ratio setting is really only the maximum achievable, but it's constantly floating between that and 1:1.

[Satanic Smurf]

Thanks for the info:)

 

Now for more questions. I guess that this isn't strictly on the thread topic of release times but it's still compressor related:

 

I was a little confused when you said that compressors didn't catch the initial transient. I thought that people used limiters (or compressors) when mastering in order to reduce the transients, thus giving more head room and allowing the volume to be increased before clipping. Is this a misconseption?

 

Compressors usually have attack times

 

Is there a way to use limiters to reduce transients without degrading the signal? Or do people mastering music accept this degredation in return for increased volume?

 

In regards to saturation reducing transients:

 

Could I try a cheap version and try running my tracks through amplitube to put a bit of drive on them to reduce transients? I don't really think that this would work but someone somewhere might have tried it and made something of it.

 

Could you recomend a good free tape saturation plug-in (VST or RTAS)? I know I could find this myself but if you know of a good one offhand it would be better.

 

OK, that's all I can come up with at the moment. I'm sure there are other questions floating around in my head but it's late in the day and work has turned my brain to mush:freak:. Sorry if these questions are incomprehensible.

[Kiwiburger]

Is this a misconseption? I believe so. The fact that a compressor has an Attack time means that all transients that occur within this Attack time will not be compressed. Destroying any chance of using a compressor as a way to "catch transients".

 

It's true that some of the very expensive limiters are fast enough to catch transients - but if you have Attack times that short, they are more likely to chew into you waveform and distort it.

 

In my opinion (probably will get shot down in flames) there is no real need to track with a compressor at all - and it only complicates the gain staging and adds noise. However - the two reasons to use a good compressor, if you know what you are doing are these:

 

1 - you like the sound it imparts - probably more of a distortion characteristic of the transformers/tubes etc

 

2 - you want to improve the Average signal level. That's Average - not Peak.

 

A good free saturation effect is Voxengo Tube Amp (don't confuse this with a Guitar Tube Amp - treat it like a tube preamp. Compared to many other saturation effects, including the famous Magneto - I prefer this. It's fairly subtle, but sounds like tubey distortion when you crank it. Other saturation effects sound like crappy fizz when you crank them.

[Zooey]

Originally posted by Kiwiburger In my opinion (probably will get shot down in flames) there is no real need to track with a compressor at all - and it only complicates the gain staging and adds noise.

 

I've been saying this for years! If you're going to track with compression, do it because you like the sound, not in the name of catching peaks.

[Satanic Smurf]

I once was blind but now I see

 

Thanks for clearing that up. While I've never used compressors while tracking (I only have the plug in version that comes with pro tools - not much use while tracking). I've always been thinking about compressors the wrong way - as a means of reducing my transients.

 

So I'm not sure if this needs a new thread or not but...

 

What are the ways of reducing transients? I know from your earlier coments that saturation is one. Are there others?

[Kiwiburger]

Magnetic tape was about the best

Transformers

Tubes

 

Digital compressors and limiters can have the advantage of 'look ahead' that hardware compressors never had. That means they have see the audio stream coming, and take evasive action before the transient arrives.

 

But - be aware that when a limiter responds to a transient, it ducks the level and then whatever follows straight afterwards gets ducked as well. That's been described as 'punching holes in your sound'.

 

You can redraw some sharp spikes in your audio.

You can automate some fader movements

[UstadKhanAli]

I rarely use a compressor for tracking. Just another thing to diddle with. I like to keep my signal path clean.

 

*IF* I had a compressor that "colors" the audio in a way that I wanted for a particular track, I'd have no problem committing to it by tracking with it, but I don't have that option.

 

A compressor can catch initial transients if it's fast enough and set for it (super fast attack). That's one way. Saturation is another. Hitting tape hard is another (dying) method . Changing the sound at the source or moving mics around can also help change the peaks a bit.

[Satanic Smurf]

So what is the method that mastering engineers use to pump up the volume?

I read a lot about music these days having no dynamic range. It's been squeezed. I always assumed that this was done with multiband compressors and limiters. Do they just bang on some saturation instead?

 

Sorry if I keep pressing the same subject but I've thought one thing for a long time (fast attack compressors reduce transient peaks and give more head room). It's sorta hard to grasp then new way of thinking.

[UstadKhanAli]

Multiband compression and limiting can get your audio a lot louder, but I think a lot of mastering guys (certainly the ones I know) are using gain optimization programs such as the Waves L1, L2, DUY Maxx, and things like that. Waves comes as either a plugin or a hardware unit.

 

And I'll pass this off to someone else because, quite frankly, I'm not entirely sure what the gain optimization is doing to achieve that and how it differs from multiband compression and limiting.

[Lee Knight]

Originally posted by Satanic Smurf ...but I've thought one thing for a long time (fast attack compressors reduce transient peaks and give more head room).

 

They do. The discussion regarding slowing the attack time to let the peaks through really has more to do with sound quality. Unfortunately getting CD's up to "competitive level" compromises sound quality. Your grasp of the scenario is correct. They limit the peaks to gain head room which in turn is used to turn the track up. If you've got an editor like Sound Forge, rip a modern CD and take a look. A picture is worth a thousand words.

[Kiwiburger]

Unfair to compare the tracking situation against modern mastered CD's. Totally different things.

 

Most modern CD's have been smashed with digital limiters with lookahead - and yes, those transients have been well and truely smashed off. But - you are also looking at the end result of probably thousands of processes and edits. Various analog and digital saturation effects have probably been used to tame a lot of transients.

 

The history of tracking with compressors comes from the days of magnetic tape. Magnetic tape smashed your dynamic range and your transients, and you didn't have to care about clipping A/D converters. You cared about Average levels, hence VU meters and tracking with slow (e.g. optical) compressors was never a worry.

 

With digital - Average doesn't cut it. Instantaneous peaks are the worry, as they will clip your converters. Very few compressors are fast enough to reduce a transient peak.

 

Or - if you have a very expensive compressor that can be set fast enough, I wouldn't like to use it as a matter of course since it would probably kill sound quality if set too agressively.

 

I doubt that anybody asking questions about compressors would own one of these compressors that are fast enough.

 

Software with lookahead is a different story. Anyone can own a fast compressor/limiter because they are free. But that's the wrong side of the A/D converters to be of any use when tracking.

[Lee Knight]

Originally posted by Kiwiburger Unfair to compare the tracking situation against modern mastered CD's. Totally different things.

 

I wasn't contridicting you kiwi. I could see the confusion about talking apples and oranges and thought I'd point out that the Smurf's not totaly off the point here.

 

Mixing and mastering are 2 different things for sure.

[Satanic Smurf]

Thanks for clearing that up guys.

 

Sorry Kiwi, I wasn't thinking about using a compressor for tracking. I've only got software compressors. Tracking with them would be sort of pointless becasue the signal has already gone through the A/D converters and like you said, you can't catch the transients anyway. There is normaly enough dynamic range in the digital world anyway. I forget how to work out the dB range for 64 bit but I remember that it was a lot.

 

So my release time question is sorted. So lets move on to the ratio question I had. I asked this before but I'll type it again so you don't have to scroll.

 

If the singal goes past the threshold (on it's way up) the imaginary slider goes down after the attack time has expired. For a given raito is the amount that the slider goes down fixed? OR if the volume were to increase further would our imaginary slider be moved down further?

 

Thanks

 

P.S. I love this board, I almost feel like I should be paying you guys for this info......but I'm not going to It's the thought that counts after all.

[Kiwiburger]

With digital audio, each bit causes a doubling in volume, which is a 6dB boost. So 64 bit internal processing = 64 x 6 = 384dB.

 

But you don't use these big numbers for improved dynamic range, it's more about avoiding rounding errors when you process the signal.

 

The dynamic range of most audio sources is usually too great and needs controlling. Tape and vinyl was around 60 - 70 dB range, and many people think that was the ultimate. Digital allows for extreme dynamic range, suitable for classical music, but generally we like to squash music more.

 

The problem is you can squash it too much.

 

The thing about a compressor is that it's gain (or reduction) is constantly floating up and down - described as the Envelope. The parameters like threshold, ratio, attack, release define the limits that this envelope complies with, but it's constantly floating in between. There are other parameters that can be used to - like Hard Knee, Soft Knee or variations in between.

 

A compressor can't make abrupt changes, otherwise the signal would be distorted. That's why they use a Sidechain signal to create a smoother envelope that varies the gain.

 

One version of hardware compressor that was/is very effective is the optical compressor. That's where the signal lights up a light bulb or LED, and that in turn varies a photo-resistor that acts like a volume control. They aren't very fast - but that's part of why they sound so good. The slowness of heating and cooling the lightbulb allows for a smooth envelope.

[UstadKhanAli]

Originally posted by Satanic Smurf If the singal goes past the threshold (on it's way up) the imaginary slider goes down after the attack time has expired. For a given raito is the amount that the slider goes down fixed? OR if the volume were to increase further would our imaginary slider be moved down further?

 

If I understand this correctly....the "imaginary slider" is your way of saying "as if someone is pulling down the fader", right?

 

If you're talking about how much the compressor compresses the audio, then given only the ratio (and not talking about other settings that can create variability with the volume - don't be smartasses here, guys ), yes, it is fixed past the threshold that you specified on the compressor.

[Kiwiburger]

Well it depends on the particular compressor. With software compressors, it's possible to be extremely accurate. You can precisely specify that if the signal exceeds the threshold, it will be precisely multiplied by the exact ratio specified.

 

Older analog stuff wasn't this accurate - you can imagine glowing lightbulbs and tubes and transformers and stuff. The parameters where more of a rough statement of intention.

 

Some of these analog compressors sounded very cool, so software models try to emulate their quirks.

 

I am fairly sure that in between the time the Threshold is exceeded, the envelope circuitry starts pulling the level down - towards the maximum reduction allowed by the Ratio parameter. This takes some time (specified by Attack) and therefore I don'

t believe an analog compressor is ever exactly multiplying all of the signal above the threshold exactly by the ratio the whole time.

 

Compression design is obviously an art - there's more going on that you would think.

[Satanic Smurf]

Originally posted by UstadKhanAli If I understand this correctly....the "imaginary slider" is your way of saying "as if someone is pulling down the fader", right?

 

I was refering to the simplification that Kiwiburger used earlier in this thread, where you imagine a compressor as an automation of a guy pulling down a fader when the singal gets too loud.

 

Now we're getting onto envelopes it would seem that the subject is a tad more complicated then I had initially assumed.

 

I've read a fair bit on compressors on the net but most of it is all the same oversimplified stuff. Do you know where I could get some detailed info on the subject? You know maths, diagrams, etc. Preferably web based (so I can slack off at work) and free ('cause they don't pay me much to slack off)

[UstadKhanAli]

Originally posted by Kiwiburger Well it depends on the particular compressor. With software compressors, it's possible to be extremely accurate. Older analog stuff wasn't this accurate - you can imagine glowing lightbulbs and tubes and transformers and stuff. The parameters where more of a rough statement of intention.

 

Well, you and I know that, but since The Smurf seems to be mostly using software compressors (plugins) and most likely isn't alternating between adjusting his Fairchild and posting on this forum, I figured I'd answer more from the perspective that he's coming from.

 

And we know that there's other variables, too, even in a software compressor, but if you're just talking about a straight compressor, sure, the ratio's fixed, even though there's plenty of ways to change that up.

 

Also, with older analog compressors, the compression ratio is "fixed" in the sense that you have a setting that is, say, 1:4. Now, that may drift a bit, do funny things within the compressor, but it's still, at the very least, an attempt to be "fixed"!

 

Sorta like pitch drift on an old analog synthesizer!! :D Or hmmmm, that sound was a little different half an hour ago.....

[Zooey]

Originally posted by UstadKhanAli Also, with older analog compressors, the compression ratio is "fixed" in the sense that you have a setting that is, say, 1:4.

 

Another common design is the fixed threshold where you vary the input gain to determine the amount of compression.

 

It's kind of counter-intuitive because you're raising the gain in order to apply gain reduction (at least it was counterintuitive for people like me whose earliest hands-on experience was with plug-ins like the Ren Comp and C1!)