i almost understand compression now...

[ikijapan]

I've thoroughly read glen's compression thing, and I've worked with compression a bit now to where I understand it.

Let me recap what I do understand correctly I think. So, if for example I'm recording a vocal and I aim to keep it right around -18 dbFS average, but then I have peaks that go all the way up to, say, -8dbFS. Then later for whatever reason I decide to compress it 5:1 with a threshold of -16 dbFS, it will lower everything above -16 dbFS at a ratio of 5db for every 1 db above that right? That part I think I understand correctly.

OK, but what I still don't understand, is, what happens to the stuff below -16 db? Does it remain completely uneffected?

I think it remains uneffected.

So, then what does make up gain do? Does that just completely increase the gain of everything by the amount you select?

Here's what I'm really wondering. What if I have a part that is too quiet in addition to the stuff that is too loud. Like what if on some parts the singer is singing too low, like at -24dbFS and then some parts at -8dbFS. I want the -8dbFS stuff to come down off course, but I also want those 1 or 2 really quiet parts to come up. What's the best way to accomplish that?

Thanks everyone!

 

[SouthSIDE Glen]

Let me recap what I do understand correctly I think. So, if for example I'm recording a vocal and I am to keep it right around -18 dbFS average, but then I have peaks that go all the way up to, say, -8dbFS. Then later for whatever reason I decide to compress it 5:1 with a threshold of -16 dbFS, it will lower everything above -16 dbFS at a ratio of 5db for every 1 db above that right? That part I think I understand correctly.

Correct!

OK, but what I still don't understand, is, what happens to the stuff below -16 db? Does it remain completely uneffected?

Theoretically, yes. There may be some unintentional coloration, especially if you use hardware compression with circuitry that colors the sound, but for all intents and purposes, there is no actual compression applied down there.

So, then what does make up gain do? Does that just completely increase the gain of everything by the amount you select?

Correct again. It's basically just a post-compression output volume control.

Here's what I'm really wondering. What if I have a part that is too quiet in addition to the stuff that is too loud. Like what if on some parts the singer is singing too low, like at -24dbFS and then some parts at -8dbFS. I want the -8dbFS stuff to come down off course, but I also want those 1 or 2 really quiet parts to come up. What's the best way to accomplish that?

There you need a dynamics processor that can do both "upward" and "downward" compression. "Standard" compressors perform only downward compression (they compress the peaks down).

Many editors such as Audition and Sound Forge come with "Dynamics Processors" that will allow you to do that (the shreen shots I use in the compression tutorial that show compression "knee" are of Audition's dynamics processor.) Those don't often necessarily have the best sound, but are servicable. For a plugin that is truely great at upwards and downwards compression, check out the Roger Nichols Digital's Dynam-izer (an older version by Elemental Audio also has a screenshot in the tutorial in that montage of pictures of compressors early on.)

G.

 

[ikijapan]

Many editors such as Audition and Sound Forge come with "Dynamics Processors" that will allow you to do that (the shreen shots I use in the compression tutorial that show compression "knee" are of Audition's dynamics processor.) Those don't often necessarily have the best sound, but are servicable. For a plugin that is truely great at upwards and downwards compression, check out the Roger Nichols Digital's Dynam-izer (an older version by Elemental Audio also has a screenshot in the tutorial in that montage of pictures of compressors early on.)

G.

OK, so at least I know I've got the other stuff figured out.

Alright, but this must be a rather normal occurance. Maybe I am just approaching it wrong. If I want to bring those particular areas up, should I perhaps go in and just up the gain in those areas? Or set the threshold for compression even lower so that more stuff is compressed and then it would be closer in volume to the lower volume areas?

Or is it common to get a seperate dynamics processor just for this purpose?

 

[RAMI]

like at -24dbFS and then some parts at -8dbFS. I want the -8dbFS stuff to come down off course, but I also want those 1 or 2 really quiet parts to come up. What's the best way to accomplish that?

Just to add something. Make sure that what you're talking about is something you HEAR, not just something you're reading on the meters. It might sound good a bit lower if it's a softer part. In that case, don't worry about what each part is clocking in at as long as it sounds good.

 

[ikijapan]

Thanks RAMI,
That is a good point. But no, I'm not just going in there are crushing everything. Just trying to figure out a few things, that's all

 

[SouthSIDE Glen]

Alright, but this must be a rather normal occurance. Maybe I am just approaching it wrong. If I want to bring those particular areas up, should I perhaps go in and just up the gain in those areas? Or set the threshold for compression even lower so that more stuff is compressed and then it would be closer in volume to the lower volume areas?

Or is it common to get a seperate dynamics processor just for this purpose?

In general the best bet, iki, is IMHO to take care of it on the individual track level and in mixing. Usually more standard compression and/or EQ on the indiviudual track - along with better mix gain levels in general - so that is will "sit better" in the mix is a more effective approach than trying to fix it after the fact by correcting too-low-level stuff in the 2mix.

But if all one has to work with is the mixdown, then yeah, upward compression is not an unusual thing for a mastering engineer to apply to make such corrections.

G.

 

[PhiloBeddoe]

Alright, but this must be a rather normal occurance. Maybe I am just approaching it wrong. If I want to bring those particular areas up, should I perhaps go in and just up the gain in those areas? Or set the threshold for compression even lower so that more stuff is compressed and then it would be closer in volume to the lower volume areas?

I think the answer is "yes".

That is, you can do either one or both. Some engineers ride faders (or draw volume curves in a DAW) to even out vocal dynamics rather than relying on compressors. This can be very tedious but takes the character of a compressor out of the equation. If you have a compressor with a sonic character that you like, then compressing with a high gain reduction may be an option.

If there are just a few words that are at a very low or very high volume, I'd probably start by manually adjusting the volume of them first. Then apply a judicious amount of compression.

 

[TexRoadkill]

Correct! There you need a dynamics processor that can do both "upward" and "downward" compression. "Standard" compressors perform only downward compression (they compress the peaks down).

Or just use a higher Ratio and turn up the Make Up gain.

If the singer is too dynamic and you can't get them to just use proper mic technique you can edit the take into two tracks. One for loud parts and the other for soft. Then adjust the volume and compression for each track. Maybe run them both to a buss with another compressor.

 

[SouthSIDE Glen]

Or just use a higher Ratio and turn up the Make Up gain.

That can bring up the low parts, but at the expense of dynamics of the louder parts. It can also raise the noise floor. If you have the (say) upper half of your dynamics the way you want but you need to bring something out of the mud, then you need to either address that in the mixing or use a multi-range compressor/expander or dynamics processor on the 2 mix.

G.

 

[ikijapan]

That can bring up the low parts, but at the expense of dynamics of the louder parts. It can also raise the noise floor. If you have the (say) upper half of your dynamics the way you want but you need to bring something out of the mud, then you need to either address that in the mixing or use a multi-range compressor/expander or dynamics processor on the 2 mix.

G.

If I go through the effort to seperate the louder passages on one track, and then the softer ones on another, is there any benefit or difference in just turning up the gain on the lower part versus using an expander?

Or is the expander more useful when there is like some really weak beginning of a word and then it goes to a normal volume or something? Just trying to understand where this would be best applied. It's just opposite of compression, right? So, the main advantage is just that the expander program would be able to react a lot faster than me going in and manually changing the gain everywhere??

Thanks guys

 

[SouthSIDE Glen]

If I go through the effort to seperate the louder passages on one track, and then the softer ones on another, is there any benefit or difference in just turning up the gain on the lower part versus using an expander?

Or is the expander more useful when there is like some really weak beginning of a word and then it goes to a normal volume or something? Just trying to understand where this would be best applied. It's just opposite of compression, right? So, the main advantage is just that the expander program would be able to react a lot faster than me going in and manually changing the gain everywhere??

Once again, the best way to handle this is to fix problems in the mix. If you have a track that has weak passages and louder ones, what you should do is level that out with volume automation on the individual track. No need to go through any effort seperating any tracks, just mix the tracks you have to sit properly in the mixdown rather than trying to fix the mixdown after the fact. It's a hell of easier that way .

G.

 

[mixsit]

Once again, the best way to handle this is to fix problems in the mix. If you have a track that has weak passages and louder ones, what you should do is level that out with volume automation on the individual track. No need to go through any effort seperating any tracks, just mix the tracks you have to sit properly in the mixdown rather than trying to fix the mixdown after the fact. It's a hell of easier that way .

G.

I second that emotion on the volume automation. Actually I'd add look at doing it at the gain or trim level (prior to the comp).
Overall the difference is unloading a lot of what's asked of the compressor(s) and ultimately going for more of a natural sound.
-Unless of course the slammed thing is in call.
Wayne

 

[SouthSIDE Glen]

I second that emotion on the volume automation. Actually I'd add look at doing it at the gain or trim level (prior to the comp).
Overall the difference is unloading a lot of what's asked of the compressor(s) and ultimately going for more of a natural sound.
-Unless of course the slammed thing is in call.
Wayne

Good points. But even if "slamming" is in order, it's usually best IMHO to slam with compression or limiting *after* the track has been basically leveled out via the automation. In other words, get the track to even "normal" levels first w/volume automation, then attack the dynamics. Those are two different problems best handled in two steps. Which sounds a lot like what you said before the slamming thing .

As far as doing it at the gain or trim level, there are several ways to trim that cat, with no single right answer. But what usually works best for me when using volume automation is to set the final trim for the track to fit the mix after the automation has been set (and after the comp, when applicable). The steps would be like this:

1.) Use automation to get the levels consistant from passage to passage (as well as ducking mistakes and whatnot)

2.) Use compression (only if desired or necessary) to tame the dynamics on the now consistant track

3.) Adjust the track trim to generally fit the track into the mix

4.) Refine automation to add any necessary mix drama

YMMV.

G.

 

[Farview]

Then later for whatever reason I decide to compress it 5:1 with a threshold of -16 dbFS, it will lower everything above -16 dbFS at a ratio of 5db for every 1 db above that right?

I could just be tired, but this is backwards.

For every 5db over the threshold, you will get 1db out the other end.

 

[ikijapan]

I could just be tired, but this is backwards.

For every 5db over the threshold, you will get 1db out the other end.

oh, yeah, i dont even understand if I typed it correctly or not, but that's what I meant. Although, now that makes me wonder, after reading how you typed it...what if it's only 2 db over the threshold...it still gets compressed at some fractional rate, right?

2db over the threshold and only 2/5 of a decibal comes out or something like that right??

 

[Farview]

oh, yeah, i dont even understand if I typed it correctly or not, but that's what I meant. Although, now that makes me wonder, after reading how you typed it...what if it's only 2 db over the threshold...it still gets compressed at some fractional rate, right?

2db over the threshold and only 2/5 of a decibal comes out or something like that right??

Correct. for every 5 that goes in, 1 comes out (5/1) If 1 db goes in, .2 db comes out.

It's a little easier to wrap your head around if you use 2/1 as an example. (well, maybe. At least the math is easier)