Mastering Compression?

[ChocolateChicken]

How do you know if you need to compress the master? Furthermore, what things should you keep in mind when compressing the master, if anything? What should be your goals with compressing the master?

I would think that compression should be handled differently when mastering than it is when mixing. I know to use a compressor on a track while I'm mixing to smooth out the waveform and reduce peaks or to make a particular instrument more punchy, but I don't know what to do with compression when mastering.

And to clarify, I do mean compression, and not limiting.

[itroitnyah]

When mastering, compression is used for the same goals. Use some light compression to smooth down the major peaks with a multiband compressor following behind it with some heavier, but still light, compression which will be smoothing out some of the bumps some more, as well as boosting the weak spots to make the frequency graph more level. But be careful to not overdo the compression, otherwise there will be major distortion. It's good to use a quicker attack time and a slow release on the compressor as well, to avoid some distortion.

[bartekko]

If a track has too much dynamics, then compress the master. short bumps are the limiter's job
Also, a limiter is a compressor with 0 attack and inf ratio

[ChocolateChicken]

When mastering, compression is used for the same goals. Use some light compression to smooth down the major peaks with a multiband compressor following behind it with some heavier, but still light, compression which will be smoothing out some of the bumps some more, as well as boosting the weak spots to make the frequency graph more level.

But if the waveform has some major peaks, doesn't that indicate a mixing issue that should be fixed in the mixing stage? And for sharp peaks, I thought that was the limiter's job. Regarding your suggestion of using multiband compression to for slightly heavier compression, could you elaborate on that some more?

If a track has too much dynamics, then compress the master.

I don't know what you mean by that; dynamics in a song is a good thing. Care to explain?

[itroitnyah]

My brain doesn't work right now. Read this. But yes, if there are major peaks they should be taken care of in the mixing stage.

As for the multiband compressor, it works simply in that it separates the frequency spectrum into 3 sections, which each have their own compressor unit on them. It's like, if you took one of Flux's songs, put a LP filter starting at around 150-200Hz. Then you took the same song, opened it in a new file, and used a bandpass filter so that only the ~200Hz-2kHz region was heard. And then you once again took the same song and used a HP filter to cut off anything below 2kHz. Then you compressed them all separately. That's what a multiband compressor does. Or is that not what you asked?

Anyways, imagine compressing each section. There would be separate peaks for each section that you would compress those sections separately in order to tame the peaks easier. Not only that, but you can apply more compression without distortion. And then you'll have more headroom.

Or am I just not getting what you're asking?

[ChocolateChicken]

Or am I just not getting what you're asking?

Actually I was simply asking you if you could talk more about what you said about your process of using multiband compression to level out the frequency spectrum instead of EQ; I already know about multiband compressors and how they work.

That guide you linked me to talked very little about compressors apart from the basics. I still don't know why a compressor would be used on a mastering chain when your track is going to be run through a limiter anyway.

Also I think I should let everyone reading this thread know that I use the Logic Pro compressors, which means I don't have to worry about distortion.

[itroitnyah]

Ok, that's easier than. Using EQ to level out the frequency spectrum just sounds ugly. Go ahead, try it for yourself. It will not work. So.... Yeah. I just apply compression to each section, and then boost or lower the gain depending, until the frequency chart looks better. Now, of course you won't get it perfectly straight, so don't worry about it. You can use EQ, but just slightly, otherwise you'll distort it too much. So overall, just play around with the gains on each section compressor

[Nine Volt]

Well, multiband compressors can split it into more than 3 bands. I've seen as many as eight and I'm sure you could do more it you want.

Distortion on compressors is an inherit quality of compressors and not limited to the exact software, it's just how compressors work. Almost all standard compressors cause distortion with low releases because that's just how they work.

As for mastering compression, I never go above a ratio of about 3:1 compression (2:1 is usually sufficient) and my threshold is usually no more than -12dB. Bottom line is you never want to overcompress your master, it sounds really bad and amateur-ish.

As for when you should compress your master, I do it lightly all the time, just to get rid of minor peaks and discrepancies that mixing won't fix without a big hassle. The point to master compression is light compression. If it sounds squashed and unnatural, you're doing it wrong.

[ChocolateChicken]

Distortion on compressors is an inherit quality of compressors and not limited to the exact software, it's just how compressors work. Almost all standard compressors cause distortion with low releases because that's just how they work.

The Logic Pro compressor never distorts the signal under any circumstances unless you actually tell it to do so.

And in regard to your advice on using light compression, I would like to know why you would need to do that, considering you would limit the master afterwards.

Also I will give your multiband compression technique a try, Itroitnyah.

I still don't understand what the purpose and goal of mastering compression is though - not multiband compression. To me, if you have sharp peaks in your master waveform, that's what the limiter is for, right? A lot of people say to use compression to smooth out the waveform on the master, but that still sounds like a job for the limiter to me. So why would compression ever be needed in the equation?

[cplbradley]

compress the shit out of it

[bartekko]

The Logic Pro compressor never distorts the signal under any circumstances unless you actually tell it to do so.

It does distort the signal you just suck at noticing it, and specifically telling it to do so creates a different kind of distortion.

I don't know what you mean by that; dynamics in a song is a good thing. Care to explain?

TOO MUCH dynamics
also if you're making EDM, you do not want the volume to change by a factor of 1000000

[Nine Volt]

Exactly, every compressor ever distorts any audio signal (with the right settings of course). Logic's isn't 'special'.

Why, in the name of all that is good and holy, would you use a limiter on a master?

And Kyoga, I only multiband compress the master, I'm just throwing out ideas. I used to compress the master though.

Also, a limiter is a kind of compressor. Just Saiyan.

[bartekko]

Why, in the name of all that is good and holy, would you use a limiter on a master?

because you make EDM

[Nine Volt]

I make EDM and don't use a limiter on my master...

[itroitnyah]

He was just joking around, 9V

[Nine Volt]

oh

[ChocolateChicken]

I find it hard to believe that anyone wouldn't have a limiter on their mastering chain, as it keeps your audio under the threshold more effectively than most other compressors. Also keeping in mind that most folks, no matter which genre, brickwall their tracks at 0dB (coughcoughAlexS.andSkrillex), and for that a limiter is comes in handy.

Although I myself don't brickwall my mixes ever, because that is the perfect way to remove punch from your song, I still utilize a limiter to keep my song from going over my specified level.

So why would compression ever be needed in a mastering chain rather than a limiter? I would think that if your intention is to "glue" your master together by compressing the peaks and increasing the loudness of the quieter materials in your song via makeup gain, then that still sounds like a job for the limiter to me. If anyone has any ideas, please do share them.

[the4thImpulse]

So why would compression ever be needed in a mastering chain rather than a limiter? I would think that if your intention is to "glue" your master together by increasing the loudness of the quieter materials in you song, then that still sounds like a job for the limiter to me. If anyone has any ideas, please do share them.

(there are multiple uses for limiters but my definition below is what I think you are using)

A limiter stops all dynamic range increases after a certain point is reached, the threshold. A signal doesn't get any louder after that point and 'stops' any peak trying to get by. It doesn't bring up the quieter like a compressor does by lowering the audios 'total dynamic range ratio' and then bringing up a makeup gain. It just stops audio from passing that threshold.

I use a compressor to squeeze the track together a bit more before letting a compressor stop any clipping (I can't explain it better than that sorry). I use it to easy the transition between a 2:1 ratio and a ∞:1 ratio.

[HMage]

If you have a quiet part of the song that's so quiet you don't hear it in portable audio like iPod headphones on a noisy subway.

A compressor with >200ms attack and >500ms release with ratio 2:0 or less usually does the job.

[ChocolateChicken]

A limiter stops all dynamic range increases after a certain point is reached, the threshold. A signal doesn't get any louder after that point and 'stops' any peak trying to get by. It doesn't bring up the quieter like a compressor does by lowering the audios 'total dynamic range ratio' and then bringing up a makeup gain. It just stops audio from passing that threshold.

But many limiters, including the one I use, have a gain control which you can use to further increase the input of the track through the limiter while keeping the signal under the threshold. So it seems like you can just raise the gain of the limiter's input to make things louder. Is this wrong?

I use a compressor to squeeze the track together a bit more before letting a compressor stop any clipping (I can't explain it better than that sorry). I use it to easy the transition between a 2:1 ratio and a ∞:1 ratio.

Don't you mean, "...before letting a limiter stop any clipping"? Otherwise, don't worry, that makes sense. You mean to say that you use a compressor to help reduce the amount of gain reduction that the limiter would have to apply after the compressor, correct?

A compressor with >200ms attack and >500ms release with ratio 2:0 or less usually does the job.

Would you care to explain why you use those particular values on your compressor? It would help me learn more about why. I agree with the ratio because mastering compression usually sounds bad when it is applied heavily. I used to think that a faster attack time, 50ms - 80ms, would work better so that it can adequately compress peaks while still preserving some of the punch of the track, considering that mastering compression sounds best when used lightly, according to what people tell me. 200ms or more sounds extremely slow to me. But you know more than I do, so I'm most likely wrong haha. As for the release time, isn't that too slow as well? Pumping is a common artifact of compression that most folks say to stay away from in mastering, although I'm not sure whether to avoid "pumping" with slower values, or to disguise the audible pumping with extremely fast release times - about 15ms or faster.

[Nine Volt]

A limiter stops all dynamic range increases after a certain point is reached, the threshold. A signal doesn't get any louder after that point and 'stops' any peak trying to get by. It doesn't bring up the quieter like a compressor does by lowering the audios 'total dynamic range ratio' and then bringing up a makeup gain. It just stops audio from passing that threshold.

But many limiters, including the one I use, have a gain control which you can use to further increase the input of the track through the limiter while keeping the signal under the threshold. So it seems like you can just raise the gain of the limiter's input to make things louder. Is this wrong

Don't do this. The 'over-limited' sound is a recognized sign of bad mastering. You're just going to end up compressing the living shit out of your track and removing all dynamics, ending up with the oh-so-famous 'brick wave'.

At least, that's what I've been told.

[ChocolateChicken]

Don't do this. The 'over-limited' sound is a recognized sign of bad mastering. You're just going to end up compressing the living shit out of your track and removing all dynamics, ending up with the oh-so-famous 'brick wave'.

At least, that's what I've been told.

Oh definitely, I agree. I never brickwall limit my songs ever. Not even with a transparent limiter because it's just lazy and sounds like shit. Somehow Skrillex is able to pull this off without any criticism. :/ Alex S, among many other musicians/mastering engineers, only brickwall limit the chorus or the drop of their songs, which I still think isn't a good thing to do.

To anyone who is new to mastering, don't brickwall limit your entire songs ever. You'll get laughed at, the listener will get tired of listening to the song, and as deadmau5 describes it, " then you've basically just got one thing. noise."

I was more so talking about just raising the gain of the limiter similar to how you would a compressor; not brickwalling.

[HMage]

Would you care to explain why you use those particular values on your compressor? It would help me learn more about why. I agree with the ratio because mastering compression usually sounds bad when it is applied heavily. I used to think that a faster attack time, 50ms - 80ms, would work better so that it can adequately compress peaks while still preserving some of the punch of the track, considering that mastering compression sounds best when used lightly, according to what people tell me. 200ms or more sounds extremely slow to me. But you know more than I do, so I'm most likely wrong haha. As for the release time, isn't that too slow as well? Pumping is a common artifact of compression that most folks say to stay away from in mastering, although I'm not sure whether to avoid "pumping" with slower values, or to disguise the audible pumping with extremely fast release times - about 15ms or faster.

When you have a compressor running so slow, it just doesn't do anything to transients, only the general volume of the section. It keeps the loud sections bit more quiet without doing anything to it's dynamics. And as a result quieter passages have less volume difference from loud ones.

You can do volume automation for this too, but sometimes compressor like this works better than automating the volume.

Somehow Skrillex is able to pull this off without any criticism. :/

Depends on how you configure your limiter and what you feed it.

[bartekko]

He was just joking around, 9V

I was not.
When not using a limiter, you can end up with many bad things
1.) Clipping when your track's gain is high (but often still way quieter than most commercially released material)
2.) A volume so low that you need an amp for your mobile devices to play it at a significant volume when listening in noisy rooms without noise-reducing headphones.
3.) When switching to your music from most other music you need to turn the volume way up, And when switching back to most other music (and I don't mean Skrillex/Guetta/Alex S. blahblahblah, but deadmau5 (who hates brickwalling) for example) you blow your ears out.
4.) When people listen to your music on youtube, the fact that yours is quieter makes it a bit less shiny, when compared to any other.

Not using a limiter is stupid and I often find that people that don't use limiters do that so they can feel better about themselves and rant about degradation of dynamics in modern music (Seriously, EDM is made for dancing, and when dancing you don't want your music to sound like beethoven. In fact, the dynamic range in some Orchestral music hurts my ears)

Besides, saying that putting a limiter on the master channel automatically makes your track a brickwallfest is a fallacy

If putting a limiter on the master sounds bad, it probably means that your limiter suck/you suck at limiters

[Nine Volt]

Alright, I'll look into proper limiting techniques.

EDIT:
Found this.

For limiting, it says that 3-4dB of gain reduction is plenty, and you should use longer attack and release times for more organic sounds. It seems to work beautifully, works nicely on a test remaster I did.