Old 04-09-2018, 09:34 AM   #1
pietro79
Human being with feelings
 
Join Date: Jun 2016
Posts: 267
Default Reacomp RMS Size

Hello!

I was playing around with the RMS Size in Reacomp and have read what it's about (http://wiki.cockos.com/wiki/index.php/RMS-size). I understand what it does operationally, under the hood so to speak, but I'm wondering what it means in practice.

In your various experiences, how/why do you use this function (in different situations, given difference sound sources)?

Why would you use RMS size versus maybe just attack and decay... I can think of some reasons maybe, but would like to hear your answers

Thanks!
pietro79 is offline   Reply With Quote
Old 04-09-2018, 11:15 AM   #2
bobobo
Human being with feelings
 
bobobo's Avatar
 
Join Date: Oct 2014
Posts: 1,356
Default

if you have the reapack and downloaded eugens stuff you will find the
envelope-based-compressor

with this fabulous script you can SEE what the compressor does in realtime. a nice tool to learn.
bobobo is offline   Reply With Quote
Old 04-09-2018, 12:04 PM   #3
ashcat_lt
Human being with feelings
 
Join Date: Dec 2012
Posts: 7,271
Default

Use longer RMS times when you want to control overall perceived level without messing too much with the relative dynamics. But remember that the RMS also effectively delays the action of the compressor, so that the gain reduction it causes may not even still be appropriate by the time it kicks in. That's where the pre-comp parameter can be really handy.

I use it all the time on about anything I want leveled. Vocals, drums, bass guitars... I even put a gentle touch of it on full mixes as part of my mastering stage.

A fairly simple example that I think illustrates the reason I use it so often:

You've got a kick drum track where the volume of each hit is relatively inconsistent one to the next. Now if you were to take a soft hit and a louder hit, normalize them to the same peak value, you would see that the shape of the waveform - the difference in level between the attack and sustain portion of the individual hits - is actually pretty close to the same. The internal dynamics of the individual hits actually are fairly consistent. They sound a lot alike once volume is compensated. But you're not going to go through and snip out each kick hit and normalize them because that would take forever.

So you grab a compressor. Using conventional techniques you basically set it so that it barely does anything on the peaks of the quieter hits but really digs in pretty good on the louder ones. Now render those two hits again. They're already a lot more closely matched as far as peak levels go, but you will notice that the one that was louder to begin with is now a lot fatter. The difference between the transient peak and the sustain has been reduced. The internal dynamics are much more different between the two hits and they will sound a lot more different from one another.

Instead you use ReaComp with a long RMS time. Now it's reacting more to the body of each hit, the overall energy of the whole hit rather than just the peak. Render that and compare the two hits and you should see that the louder hit doesn't get squished near as much, the whole hit is turned down and the two look and sound a lot more alike again.

Except of course that by the time the long RMS time has added up to enough to make the compressor start to turn down, the hit that "triggered" it is probably over. And that's where pre-comp comes in. I've found that setting that to about half of where RMS works just about right for most things. This of course limits how long of an RMS time you can use since precomp tops out at 250ms. I wrote a JS "clone" of ReaComp pretty much just so I could have longer lookahead times, but TBH I don't too often find it necessary.

I have a ReaComp preset called "lookahead leveler" that I slap on all kinds of things. Precomp 250, RMS 500, attack and release at 0 because they're essentially irrelevant to what we're doing. For vocals I might set the ratio as high as 6:1, though if I have to go that far it's probably better to get in and automate the Take or PreFX Volume envelope first. On individual drums, bass, cleanish guitars maybe 2:1 or 3:1. On a mix bus I won't ever go above 1.1:1, just a little squishy glue, but on a drum bus I might go as far as 1.5 or so.

It's seriously almost like cheating. It's a great way to level individual tracks and an extremely transparent way to inflate your perceived level without murdering your transients because it kind of dips around those transient but leaves their basic shape intact.

Course precomp adds latency. That's all compensated for most of the time, but it can't work in realtime - like on live inputs. In those cases you're often better off "making due" with more traditional techniques. Long RMS can still help a bit though if you've got something that goes from quiet to loud and stays there for a while. Like a guitarist who hasn't got his boost pedal set quite right. It won't catch it right away, but it will ride the level down in much the same way a human sound guy might if suddenly surprised by something like that.

I use it to keep my robot drums from just pounding away as loud as they can any time I turn on a pedal. They follow my dynamics reasonably well, but when I get really loud, things can be a bit too much. So RMS compress the guitar signal that us modulating the MIDI velocities. Now if I go from quiet to loud, the drum will hit hard once or twice - fairly appropriately emphasizing the dynamic changes. If I stay loud, though, it will calm back down a bit until and unless something changes. Thats not exactly typical usage, but maybe illustrates a bit more how it works?

Edit - You really should just play with it on different sources, but here's one more example:

Consider a full mix of one of those soft verse/loud chorus kinda songs. The chorus just gets a bit too much louder - in both peak and average or perceived level. You can try to squash down the peaks of the chorus. Then it's average levels stays about the same but it's less dynamic than it "should be" in extreme cases it starts to sound unnatural and unless you're really careful, it could actually sound distorted.

And RMS compressor essentially just turns the whole shit down but leaves the dynamics intact. Without lookahead, it'll miss the first hit or two. They'll still be a lot louder and you'll get that "transitional impact" and then things settle back down a little. But that can feel a lot like somebody when "Holy shit!" and reached for the volume knob. With some precomp, it will start turning down at the end of the verse so that by the time it gets loud, it's already where it needs to be. This might reduce that "transistional impact", because it doesn't get quite as much louder, but the louder chorus maintains more of its own natural dynamic impact, and it usually works a heck of a lot better. You might say "Volume Automation..." and you'd be right, except who's got time for that? This "trick" works so well that it's almost a waste of time to not use it. Slap it on there and use your time for something more important. Plus, it does some subtle things that are almost indescribably in the way instruments interact with one another and the space around them. At the extreme, it's the difference between fucking on a waterbed as opposed to a concrete floor.

Last edited by ashcat_lt; 04-09-2018 at 12:27 PM.
ashcat_lt is online now   Reply With Quote
Old 04-09-2018, 12:44 PM   #4
pietro79
Human being with feelings
 
Join Date: Jun 2016
Posts: 267
Default

Thank you bobobo & ashcat_lt

ashcat_lt, thank you for taking the time to give a longer explanation. I will certainly need to fiddle around more to more intuitively understand the meaning of your examples.

I was just about to ask you to elaborate the difference between using RMS size to achieve what you described versus using attack and release times--but then it suddenly sunk in...I think.

"Instead you use ReaComp with a long RMS time. Now it's reacting more to the body of each hit, the overall energy of the whole hit rather than just the peak. Render that and compare the two hits and you should see that the louder hit doesn't get squished near as much, the whole hit is turned down and the two look and sound a lot more alike again."

So, let me recap in my own ineloquent words:

the RMS size determines *how long* of a moment to average (a window which is essentially a length of time). The compressor then uses this window to find the average amplitude of that moment, and the compressor does its thing to that moment as a whole... *without* changing the difference within this window between "loudest" and "softest" (highest and lowest values of amplitude) part of the signal (and in that sense kind of acts like normalization within this micro-temporal window)

I like your description better, but is mine totally off?
pietro79 is offline   Reply With Quote
Old 04-09-2018, 02:47 PM   #5
ashcat_lt
Human being with feelings
 
Join Date: Dec 2012
Posts: 7,271
Default

I think you've got it pretty well. Important to remember though that that "window" keeps rolling sample-to-sample. It's not like snipping the thing into 500ms sections and adjusting them separately. There are different ways to calculate that window. Whether it is an actual window or just basically a lowpass filter ultimately doesn't matter much. Gain reduction right now depends on the last 500ms and gain reduction on the next sample is 500ms back from that. So the gain reduction is really smooth.

But yeah it reacts more to longer term volume changes and lets quicker blips get through.

More traditional compression is reacting (almost) instantaneously to the actual sample levels, but then the A/R slows down the actual movement. It's a fine but important distinction that I think is better illustrated if you think about a noise gate. If the gate is looking at a sample-by-sample level, it's going to want to open any time a little blip goes over Threshold. It'll start opening, but that'll take a little while but then it should really be closed but then THAT takes a little while and you get that weird chattering thing happening. If the gate was reacting to an RMS average, it wouldn't even bother trying to open until it gets loud and stays there for a while.
ashcat_lt is online now   Reply With Quote
Old 04-10-2018, 05:52 AM   #6
Eliseat
Human being with feelings
 
Eliseat's Avatar
 
Join Date: Mar 2018
Location: Cologne
Posts: 1,362
Default

ashcat_lt

If I' am not wrong, you could also copy the needed audio track, set it up in the arrangement to -250ms and use it as a side chain trigger to avoid to much latency. (No pre comp needed!) I know latency gets compensated very well in Reaper, but all your visual signals are out of timing then. At least on my laptop.
Eliseat is offline   Reply With Quote
Old 04-10-2018, 10:28 AM   #7
ashcat_lt
Human being with feelings
 
Join Date: Dec 2012
Posts: 7,271
Default

Eliseat - You're not wrong about that. It's a little less convenient and obviously can't help on a live input, but it will work. In a mix, the little bit of latency doesn't normally bother me. If I did more realtime automation (with a hardware controller or whatever) it might be a bit more trouble.
ashcat_lt is online now   Reply With Quote
Old 04-10-2018, 04:50 PM   #8
Tod
Human being with feelings
 
Tod's Avatar
 
Join Date: Jan 2010
Location: Kalispell
Posts: 14,745
Default

Quote:
Originally Posted by ashcat_lt View Post
Use longer RMS times when you want to control overall perceived level without messing too much with the relative dynamics. But remember that the RMS also effectively delays the action of the compressor, so that the gain reduction it causes may not even still be appropriate by the time it kicks in. That's where the pre-comp parameter can be really handy.
Wow ashcat, that's an incredible post. I'm going to copy/paste it for later, there's no way I can comprehend all of it right now.
Tod is offline   Reply With Quote
Old 04-15-2018, 12:37 PM   #9
pietro79
Human being with feelings
 
Join Date: Jun 2016
Posts: 267
Default

Thanks again everyone!

As I said, these sorts of functions take a bit of actual practice to grasp at a more intuitive level, but you've helped me start to get it. Much appreciated!
pietro79 is offline   Reply With Quote
Old 05-26-2018, 05:00 AM   #10
Valle
Human being with feelings
 
Valle's Avatar
 
Join Date: Nov 2007
Location: Sweden
Posts: 808
Default

Quote:
Originally Posted by ashcat_lt View Post
Use longer RMS times when you want to control overall perceived level without messing too much with the relative dynamics. But remember that the RMS also effectively delays the action of the compressor, so that the gain reduction it causes may not even still be appropriate by the time it kicks in. That's where the pre-comp parameter can be really handy.

I use it all the time on about anything I want leveled. Vocals, drums, bass guitars... I even put a gentle touch of it on full mixes as part of my mastering stage.

...
This must be the best tip I have ever read, all categories!

Thank you so much for this!
__________________
Valenzia Vision
Valle is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 11:48 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.