Compression

PaulBrewer

'If the gain reduction doesn’t return to zero several times per bar, you’re compressing too much'

Discuss ?


http://www.homestudiocorner.com/2011/06/13/over-compression/?awt_l=5Jfpw&awt_m=IgyLFggPqqazuu

trackmixstudio

So there is a right and wrong way now?:rolleyes:
I just tweak until it sounds good.

woodsdenis

PaulBrewer wrote: »

'If the gain reduction doesn’t return to zero several times per bar, you’re compressing too much'

Discuss ?


http://www.homestudiocorner.com/2011/06/13/over-compression/?awt_l=5Jfpw&awt_m=IgyLFggPqqazuu

No Phil Collins drum sound then or Hugh Padgham ?

drumdrum

PaulBrewer wrote: »

'If the gain reduction doesn’t return to zero several times per bar, you’re compressing too much'

Discuss ?


http://www.homestudiocorner.com/2011/06/13/over-compression/?awt_l=5Jfpw&awt_m=IgyLFggPqqazuu

Hmm.....depends on whether you re talking about peak or RMS compression on a track.
If its for peaks, then I agree, the track should return to zero once the peak have been caught. That said, if you are peaking a lot within a small range that its more an average dynamic range, then I say that you are tracking too hot and to turn the input down a bit. So your range that was once between -2db and +2db, is now between -9db and -5db.

If its RMS value, then surely the compressor will likely have a state of "always on", with variation in the reduction gain characterizing the sound.

The article does have a point though....compression is overused. If you have an inconistant player thats tying your hands, then thats one thing, but knowingly compressing the **** out of things that dont need it is madness!

I mean, would you compress a distorted guitar signal seeing as how its so compressed anyways coming out from the amp?
(I'm actually curious about this one and other peoples thoughs....even though it comes across a bit "matter of fact"-ish! )

Paolo_M

Shouldn't this read "If the gain reduction doesn’t return to zero several times per bar when peak levelling, you’re compressing too much"?

Isn't smashing the **** out of a track as an effect perfectly valid?
What about parallel compression where you smash something heavily?

These blanket statements always baffle me, with music it's all about context and what works best.
It's the reason why some people make good music producers, and other never will with any amount of edumacation.


dumdrum, I'd use multi-band compression to catch any cab resonances present. It's a trick I picked over at the Sneap forums and I find it works very well, streets ahead of anything I was ever able to get with EQ. It's like a de-esser for the lows/low midz.

drumdrum

Paolo_M wrote: »

dumdrum, I'd use multi-band compression to catch any cab resonances present. It's a trick I picked over at the Sneap forums and I find it works very well, streets ahead of anything I was ever able to get with EQ. It's like a de-esser for the lows/low midz.

Yeah I saw something like that mentioned in a few places on the net. Havent tried it myself yet though...

So how would you set your frequency bands, and how much would you recommend setting the threshold and ratio at? (obviously this varies from person to person, but any ballpark figures?)

Is the idea just to get the lows and low mids to be brought forth a bit more?
I tend to high pass anywhere from around 100 - 150Hz to give room for the kick drum and bass, so my lows are already fairly altered to start with. I low pass usually from about 8.5KHz to 10kHz to leave the top end for the cymbals...
Lately I've been parallel compressing them to decent results.

Compression with vocals can be amazing though, especially if your vocalists mic technique is a bit ....well....crap!

Paolo_M

drumdrum wrote: »

Yeah I saw something like that mentioned in a few places on the net. Havent tried it myself yet though...

So how would you set your frequency bands, and how much would you recommend setting the threshold and ratio at? (obviously this varies from person to person, but any ballpark figures?)

Is the idea just to get the lows and low mids to be brought forth a bit more?
I tend to high pass anywhere from around 100 - 150Hz to give room for the kick drum and bass, so my lows are already fairly altered to start with. I low pass usually from about 8.5KHz to 10kHz to leave the top end for the cymbals...
Lately I've been parallel compressing them to decent results.

Compression with vocals can be amazing though, especially if your vocalists mic technique is a bit ....well....crap!

My preference is to high pass to around 80Hz, depending on the track.
I'll then set up the multi-band to compress only from 300Hz ish downwards, depending on the size of the room (Sneap goes from around 200Hz but I've had to go up to 350Hz as I'm recording in a box bedroom I guess), so that it catches resonances and compresses them hard.
Any other bands should be deactivated, distorted geetar is compressed enough already.
I'll usually pick a palm muted section of the track for this as it brings out any chunky gunk that's happening the most.
I'll set the thresh-hold to compress only on big chugs, set the ratio pretty aggressive (6-8), set the attack pretty quick and the release pretty quick too, depending on the music and tempo.
So for good parts of the song the compressor wont engage at all, just like a de-esser mightn't engage at all if your song has no sibilance.

It's about controlling the lows of the guitar, not getting more or less of them. The same theory you'd use for bass I guess.

After that you can EQ more or less (as the music requires) without worrying about having a complete sh*t fest happening in your lows/low midz.

krd

I did something really thick.

I recorded something back in December. A live jam on a Roland Groovebox - about twenty something minutes long. It turned out really well. But then I took the file and maxed the volume on it (using comp and limiting). I played it back today......and it's way too loud. It's just too loud by a few notches - which will cause problems on most things it will play back on. Play it on monitors, it will just sound loud. Normal speakers, it will sound too much - crackle and distort on the subs.

I hope I have the original file

If I don't I have to figure out a way to fix the MP3.

DamagedTrax

krd wrote: »

If I don't I have to figure out a way to fix the MP3.

did you try expansion? if its not totally slammed it can sometimes work. ive gotten better results using M/Band but its not a perfect solution at all.

krd

DamagedTrax wrote: »

did you try expansion? if its not totally slammed it can sometimes work. ive gotten better results using M/Band but its not a perfect solution at all.

I haven't tried anything with it yet. I hope I find the original.

I just went that little bit too loud with it. It sounds right - just it fills too much of the dynamic range (I might be able to fix the MP3 by giving it a little headroom) it's not actually hitting the red - though I know if it's played back through ordinary speakers, it's just going to be way to loud through the bass - you'd need to turn it down sharpish.

Loudness is a funny thing. I'm realising things like timbre can play a major part in how loud something sounds. I'm pretty sure compression can sometimes make things sound quieter - and you're losing something by using it.

EMPI

Hey All,
Compression is great but you really have to know when not to actually use it. So many engineers and producers use far too many effects and compression. They worst is when they take something like a classical voice and compress the hell out of it. There goes the actual sense of every word sang! :rolleyes: I recorded a soprano who has a masters in opera. She said every studio she recorded in compressed her vocals and she didn't like it. When I first recorded her I used various tricks without compression to make her vocals sound great even when it was acapella. The thing is it is more time consuming this way but it's worth the results! I think engineers and producers are returning to the oldschool approach more and more and the digital age is so far gone that it's caving in on itself. Then as we always see it comes full circle.

Regards,
EMPI

Hayte

Article is very wrong. It uses the example of 10dB constant gain reduction but the effect of a compressor is a ratio of peak input over threshold to peak output over threshold. The amount of gain reduction is constantly changing based on the peak input signal over threshold.

If you have hard knee, absolute (not RMS) peak level detection and 0ms attack/release then this is the fastest you can make auto gain reduction track the input signal. It will produce artifacts if there is a sudden, large amount of gain reduction in a short space of time. This may or may not be desirable depending on what you want to do.

If you want less sensitive tracking to avoid artifacts there are loads of ways to do that - use an RMS compressor (and increase RMS window size accordingly), use an attack or release envelope to delay the onset and relaxation of gain reduction and/or use soft knee (gain reduction is a sliding scale that begins x dB below threshold and reaches maximum when the input signal exceeds the threshold).

In the case of hard knee compression, nothing is affected below threshold, so the idea that 10dB of gain reduction from a compressor is equivalent to turning down the output by 10dB is flat wrong. It isn't even true for a soft knee compressor.

If gain reduction is constantly happening then you are constantly transforming the signal (above the threshold).

johnnylakes

''Compression is for kids'' - Bruce Swedien

Hayte

Not to detract from the man's formidable achievements but the GS thread where that quote comes from is a rambling mess. I think he may have been drunk when he posted it.

madtheory

Hayte, I don't understand what you mean- dB is a ratio?

Regardless, I think compression is over used. Limiters are usually more appropriate because they have more benign side effects.

krd

madtheory wrote: »

Hayte, I don't understand what you mean- dB is a ratio?

Regardless, I think compression is over used. Limiters are usually more appropriate because they have more benign side effects.

Really, I think it comes down to your ears. My ears are still developing. I still have a fair bit to go, and I haven't been seriously getting down to it recently - which I think may have helped.

But for example. Last year I downloaded a cymbals sample pack. At the time, when I listened to the samples, they sounded fine - but of course I couldn't get them to work - something didn't sound right about them but I couldn't understand what was wrong with them. I listened to them again just the other day - and about a few milliseconds into each sample, they suddenly become horribly distorted. It's so obvious to me now, but last year there was no way I could hear that.

Definitely, I think people burn their mixes with misapplied compression - digital is pretty unforgiving.


Your ears can play tricks on you. The first time I took a serious stab at compression, it was a complete disaster - but I didn't hear it until the next day - like my ears were doing a weird correction on the sound, when I was recording.

Jagle

madtheory wrote: »

Regardless, I think compression is over used. Limiters are usually more appropriate because they have more benign side effects.

statement does not compute
a limiter is a compressor

mkegvn

Compression is a black art if ever there was one. The hours I've lost!...

....The hours I loose, I should say.

Hayte

madtheory wrote: »

Hayte, I don't understand what you mean- dB is a ratio?

Regardless, I think compression is over used. Limiters are usually more appropriate because they have more benign side effects.

Nah, the amount of gain reduction that a compressor applies is a ratio of input gain over threshold to output gain over threshold.

So if you have 9:1 compression ratio then it means that if the peak input is 9dB above the threshold then the peak output is suppressed to 1dB above the threshold. Similarly, if the input peaks 18dB above threshold then the output peak will be suppressed to 2dB above threshold. You don't have to use integer values but its easier for illustrative purposes. The threshold is a fixed and arbitrary point above which gain reduction occurs. The input signal will vary alot in amplitude and you can get a (very) rough sense of how much by watching your peaking meters bob up and down. The amount of gain reduction you get from a compressor varies with the input above threshold.

A compressor can be made to behave like a limiter at very high compression ratios and high thresholds. You can absolutely destroy any sound by lowering a limiter's threshold because it instantly applies massive and indiscriminate gain reduction if the input goes over threshold. That makes sense because it is by design meant to suppress transients that would destroy your soundsystem (and your ears).

mkegvn

Jagle wrote: »

statement does not compute
a limiter is a compressor

Limiting is a type of compression. They're not the same.

Jagle

mkegvn wrote: »

Limiting is a type of compression. They're not the same.

there the same to the point that its a comp with different values, using a comp with limiter values will limit

madtheory

Plus a limiter has very fast attack and automatic release time. So it's only kind of a compressor. If it's just a reduction of dynamic range you need, then a limiter is the way to go. Compressors usually give you a lot more control so you can shape the "envelope".

Hayte, right, gotcha, that's much clearer. But what about using the GR meter?

Hayte wrote: »

it instantly applies massive and indiscriminate gain reduction if the input goes over threshold.

Not sure what you mean by indiscriminate- the "discrimination", such as it is, depends on where you set the threshold.

johnnylakes

Hayte wrote: »

Not to detract from the man's formidable achievements but the GS thread where that quote comes from is a rambling mess. I think he may have been drunk when he posted it.

He said it in a video interview with Dave Pensado (Pensados Place) . I was just messing anyway! Easy for him to say with the gear and musicians etc he had at his disposal! Compression is a can of worms anyway. Lately I have been trying to use it as sparingly as possible, if you can get the source right really is the big thing. I think everyone goes through a phase of slapping it on everything when starting off. Less is more and all that.

Hayte

Jagle is correct.

A compressor essentially becomes a limiter at very high compression ratios and very short attack/release times.

Its not like you even need to be specific about attack times since some compressors do stupid attack times. Crane Song STC-8 claims less than 1 microsecond attack which is hilarious because oh the distortion. You can also have soft limiting and some limiters don't have compression ratios as high as some compressors even. So theres a load of overlap and theres no clear boundary where a compressor ends and a limiter begins. If you are using a compressor to catch transients with a fast envelope, high threshold and high ratio then you are limiting. Plain and simple.

johnnylakes wrote: »

He said it in a video interview with Dave Pensado (Pensados Place) . I was just messing anyway! Easy for him to say with the gear and musicians etc he had at his disposal! Compression is a can of worms anyway. Lately I have been trying to use it as sparingly as possible, if you can get the source right really is the big thing. I think everyone goes through a phase of slapping it on everything when starting off. Less is more and all that.

I wonder where it came from first? My favourite Bruce quote is when he said that a chimpanzie could record Michael (Jackson). He was literally that good at performing that he makes other people feel good about their work even though they pretty much have to do jack s**t.

Really puts things into perspective when you consider all of this is just shop talk really. I still don't like shovelling bulls**t into people's heads because theres nothing more saddening on the internet than someone asking a technical question, and someone else giving them a technically wrong answer. Then they go off and get insane ideas in their head. Years later you see them on Gearslutz or some other fringe internet society. Not even grade 1 standard with their first instrument but with 50k worth of gear. Not even my enemies deserve that.

krd

Jagle wrote: »

statement does not compute
a limiter is a compressor

You think of each differently.

The limiter is to limit the peak volume - a bit like giving your track a crew cut, across the top. (that's if you do brick wall limiting on a finished track - to maximise the volume. )

And then you think of the compressor to lift quieter sounds to a higher level.

Is compression is a lot trickier - you'd need a lot of experience to be able to visualise the lift and drop curve - or just to have the Voodoo, to "know" what will happen. I don't have no Voodoo.

madtheory

krd wrote: »

And then you think of the compressor to lift quieter sounds to a higher level.

Aargh! You might want to revise that, especially in light of hayte's last post...

Anima

It's correct though. A compressor is used to lift the average RMS of a signal by sacrificing dynamic range (or by reducing the difference between the highest and the lowest points of the signal).

With a higher average RMS, the compressed signal will appear louder to the ear overall even though it doesn't reach the same peak but it will have less attack due to the peak reduction. Which is why compression can ruin decent sounds as you usually want to keep dynamic range for the most part.

That's my understanding, it is still damn difficult to get it sounding right though and even harder with multi-band compression.

madtheory

Yes you're right (except that "average RMS" is a contradiction) but that's not what he said!

If nothing else, all this shows that compression is a minefield.

Never saw the point of multiband myself. If a mix "needs" it, then what's needed is a remix. And it makes no sense on individual instruments. No wonder you find it harder, it's using an elephant gun to shoot a mouse. Those things were originally designed for US radio stations to create a "signature" sound in an effort to be both louder and different on the crowded FM wavelength. Then there was a craze for them in mastering. Now the myth has been established. Although they do have some creative uses alright.

Anima

You're right I shouldn't have said average, just RMS. "Lifting the quieter sounds" is what it does though but it's a side effect of what the actual compressor does. By killing the peaks, you create room to turn the gain up afterwards (without clipping) which of course means the quieter elements get louder (including noise), no?

I wouldn't use multi-band compression on a narrow bandwidth signal obviously but in electronic music there can be synth sounds that can be quite wide.

It's great for stabilising the low end, letting the mids move and then keeping the high end tight. Having said that I usually avoid compression if I can because more often it sounds better to achieve a similar end by a different means, like you say.

madtheory

I think it's important to note that "being able to raise the level of previously quite sound" is a side effect. The actual compression is done on the peaks. And that's where it can hurt the sound the most.

I make most of my own music electronically, and never had a need for a multiband. I would think that appropriate choice of sound/ good programming and good arranging would be a better solution.

Going through each control one by one, by setting the others at extremes, is the best way to learn what the compressor is doing. So for example, starting at the beginning- set attack in the middle, release on fast or auto, threshold very low and ratio very high. Then move the attack control and get a feel for it. It's an old technique, but Stav describes it well in Mixing with your Mind.

Anima

Yeah that's good advice.

I found this VST below very useful to see how it's working as well. It's a Limiter/Compressor that comes with FlStudio. The white line in the background is the amount of compression being applied, the purple is the original input signal and the grey/beige waveform is the output signal, the green solid line is the threshold.

It's very useful to see how the different parameters are affecting the signal in real time.

madtheory

That is the coolest GR meter ever!

Seziertisch

I've been using this quite a bit recently.

http://www.irishaircompressors.com/files/dental2.pdf

Getting really good results, I a/bed it with the plug-in version and there was just no comparison.

Seziertisch

Otherwise, this is worth a read

http://www.mbrauer.com/articles/tapeop.asp?pp=1

krd

Although, I am an amateur, who has trouble distinguishing arses from elbows, and Brasso from dogsh1t.

I believe now, that overall compression, is a bad thing, because it's too imprecise. And better results can be achieved through going around the long way on the sounds.

Imagine it this way - you're like a painter. In a smock, and with a berry. And you gently touch up the levels - with your brush (the pencil tool) - that would be the artistic think to do. Or, you could be a painter "and" decorator, from Navan, with his roller, his 80 a day fag habit, and drink problem, and his roller - givin' it an f'ing job. ........In magnolia.

You have to decide.

Are you going to be a painter.

Or

A painter and decorator.

What's the aural equivalent of wall paper?

i57dwun4yb1pt8

krd wrote: »

You have to decide.

Are you going to be a painter.

Or

A painter and decorator.

What's the aural equivalent of wall paper?

this

http://www.youtube.com/watch?v=gcF9ICgLqi4

Hayte

Hrmmm, I'm not sure I understand the analogy but compression is very precise. You determine the compression ratio and the amplitude at which it starts working. Even if you set the threshold as low as it will go (full scale gain reduction), it will track the input far quicker and more accurately than if you were to ride the channel volume fader.

krd

Hayte wrote: »

Hrmmm, I'm not sure I understand the analogy but compression is very precise. You determine the compression ratio and the amplitude at which it starts working. Even if you set the threshold as low as it will go (full scale gain reduction), it will track the input far quicker and more accurately than if you were to ride the channel volume fader.

Ratio? Amplitude? Threshold?

Are you some kind of mathematician, physicist, or something.

You know it might be the wrong way to think about things. Like, "Hey, this melody line sounds a little weak and grooveless - why don't I whack up the magic sauce, and make it louder"

You know, I think compression could be the equivalent of one of those penis pump things. Yes, it might give you a bigger cock - but it will be no short cut to making you a master love maker. I want you to think about that metaphor a little. Pause........and reflect.

madtheory

krd wrote: »

Ratio? Amplitude? Threshold?

Are you some kind of mathematician, physicist, or something.

In fairness, the job is sound engineering. Being mad and creative is fine if it's your own compositions, but if you're the person hired to record and mix it, then you damn well should be an engineer. And being technically adept does not preclude creativity. It just takes a bit more effort.

krd

madtheory wrote: »

In fairness, the job is sound engineering. Being mad and creative is fine if it's your own compositions, but if you're the person hired to record and mix it, then you damn well should be an engineer. And being technically adept does not preclude creativity. It just takes a bit more effort.

Technical adeptness is also in knowing when not to use something.

And I'm not even talking in terms of creativity.

And you don't call someone who makes handmade violins, flutes, lutes, and guitars for a living, an engineer. You call them a craftsman. And engineer is a Big Ignorant F****, who digs the road up with a shovel.

There is a saying about carpenters...that a true carpenter never uses a hammer nor nails.

Hayte

krd wrote:

Ratio? Amplitude? Threshold?

Are you some kind of mathematician, physicist, or something.

You know it might be the wrong way to think about things. Like, "Hey, this melody line sounds a little weak and grooveless - why don't I whack up the magic sauce, and make it louder"

You know, I think compression could be the equivalent of one of those penis pump things. Yes, it might give you a bigger cock - but it will be no short cut to making you a master love maker. I want you to think about that metaphor a little. Pause........and reflect.

I don't know why you are laying into me. Ratio, Amplitude and Threshold are terms that you see on all compressors so if you understand what they do, then you can use all compressors. It helps to know how they work if you want to have any control over what you are doing, otherwise everything is just accidental. Sometimes you can have a happy accident and its all good, other times you'll have a complete disaster and you won't know how to fix it. Shrugs.

You don't have to learn how a comp works if you don't want to and I'm certainly not forcing you.

krd

Hayte wrote: »

I don't know why you are laying into me. Ratio, Amplitude and Threshold are terms that you see on all compressors so if you understand what they do, then you can use all compressors. It helps to know how they work if you want to have any control over what you are doing, otherwise everything is just accidental. Sometimes you can have a happy accident and its all good, other times you'll have a complete disaster and you won't know how to fix it. Shrugs.

You don't have to learn how a comp works if you don't want to and I'm certainly not forcing you.

It's nothing personal......it's not meant to be personal...it's more of
a philosophical discussion.

Anyway... I'm listening to the radio - Phanthom - And I'm wondering, their compression: are they using some kind of groove compression to give the music an extra pulse?

madtheory

krd wrote: »

Technical adeptness is also in knowing when not to use something.

No one said otherwise.

krd wrote: »

And you don't call someone who makes handmade violins, flutes, lutes, and guitars for a living, an engineer. You call them a craftsman. And engineer is a Big Ignorant F****, who digs the road up with a shovel.

I think you'll find that the better luthiers know "a thing or two" about maths and physics, but that's beside the point. We're talking about using compression, in the course of working as a sound engineer.

You're making the typical incorrect generalisation- engineer applies to more than just civil engineering (which is what I assume you're describing so colourfully). Here is how Audio Engineering is defined on the Surrey Tonmeister course:
"The study of technical aspects of audio such as electronics, acoustics and signal processing."
http://www.surrey.ac.uk/msr/study/ug/bcourse/index.htm