How do you null test your limiter

This forum is for discussing Reason. Questions, answers, ideas, and opinions... all apply.
Post Reply
User avatar
Billy+
Posts: 4157
Joined: 09 Dec 2016

08 Sep 2020

So I was just watching The future of mastering, and the speaker used a technique to show the results of the limiter called a null test and that interested me very much but I really don't get how he achieved it, so can anyone elaborate how this would be done inside Reason.


User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

08 Sep 2020

I’ll start by pointing out I’m by no means an expert, but I have done a decent amount of work playing around with using phase cancellation to produce different types of effects.

the way you could get it to work in Reason would be to take an audio track with no processing on it, and then create a parallel track. they have to have all the exact same settings. then route them both to a new output bus. flip the polarity on one of the two parallel tracks (not both—this is the only setting that should differ). they should completely phase cancel, and you shouldn’t hear anything.

now add your limiter to BOTH tracks. you can mute one for now. with the one still unmuted, set the limiter how you like it. when you’re happy with it, copy and paste that setting to the limiter on the muted track.

now you can unmute the muted track. the result should still be silence. if not, you’ve either done something wrong, or the plugin is imparting some element of randomness, which will throw off the result. any plugins that won’t treat the same audio exactly the same way every time they play back will result in a non-silent signal at this point, even if the settings are the same.

assuming it’s silent, you can now start tweaking settings on one of the limiters. as you do, you should start to hear only the difference between the two.
I write good music for good people

https://slowrobot.bandcamp.com/

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

08 Sep 2020

I just saw that today too, weird…
But yea, SUPER easy to do in Reason without using any plugins/REs.
Just either duplicate your audio clip or create a parallel channel and add a limiter on one channel, or create multiple channels if you have more than one limiter to compare. Then invert the polarity ("INV" at the top of the mixer channel strip) on the original so you can then audition different compressors by muting/unmuting their channels.
The other important thing to do that he didn't mention was to level match the parts below the threshold. This must be done each time you adjust the threshold, so the "trick" would be to choose an amount of gain reduction and loop a 4-8 bar section so you can hear the same audio compared. So you want an audio source that peaks around 0 dBFS without going over, then if you set the Limiter to -6 dB, you need to also subtract 6 dB from that channel (since the limiter boosts the signal according to how you set the threshold). You want the source audio to be close to 0 dBFS because if it's not at least higher than -6 dB you won't hear anything because there won't be any "limiting".

The other important issue is delay, which almost all mastering limiters use. If this exceeds the limit of the built in delay compensation, you can't do this test. For example, the Ozone RE exceeds the limit at one setting, IRC II, so you can't test that setting. Surprisingly, Ozone Elements VST does not exceed the limit when set to IRC II, go figure.
(I was writing this when guitfnky posted - both approaches will work - you can use their approach to test any algorithm that exceeds the compensation delay amount)
Selig Audio, LLC

User avatar
Billy+
Posts: 4157
Joined: 09 Dec 2016

08 Sep 2020

Thanks all, very interesting trick

User avatar
Billy+
Posts: 4157
Joined: 09 Dec 2016

08 Sep 2020

Snagged the target levels if anyone wants to know them
IMG_2154.JPG
IMG_2154.JPG (102.63 KiB) Viewed 1555 times

PhillipOrdonez
Posts: 3732
Joined: 20 Oct 2017
Location: Norway
Contact:

08 Sep 2020

Billy wrote:
08 Sep 2020
Snagged the target levels if anyone wants to know them
IMG_2154.JPG
Thanks for the video link. Sound like an AES conference I watched some years back. Super interesting stuff.

As for the targets, unless you're producing acoustic music do not try to master your music to those targets...

User avatar
Billy+
Posts: 4157
Joined: 09 Dec 2016

09 Sep 2020

PhillipOrdonez wrote:
08 Sep 2020
As for the targets, unless you're producing acoustic music do not try to master your music to those targets...
Yeah I'm glad you pointed that out but that does make me wonder why they are listed in the presentation?

What general targets would you suggest?

Mine are as follows

Code: Select all

Balanced Mix
INT - Integrated LUFS   -13
 ST - Short Term LUFS   -11

dBTP - True Peak        -3 dBTP 
  dB - Peak             -4 dB

DR - Dynamic Range      10 DR
LU - Loudness            3 LU

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

09 Sep 2020

Billy wrote:
09 Sep 2020
PhillipOrdonez wrote:
08 Sep 2020
As for the targets, unless you're producing acoustic music do not try to master your music to those targets...
Yeah I'm glad you pointed that out but that does make me wonder why they are listed in the presentation?

What general targets would you suggest?

Mine are as follows

Code: Select all

Balanced Mix
INT - Integrated LUFS   -13
 ST - Short Term LUFS   -11

dBTP - True Peak        -3 dBTP 
  dB - Peak             -4 dB

DR - Dynamic Range      10 DR
LU - Loudness            3 LU
I think these targets aren’t really worth spending too much time worrying about unless you’re only looking to hit one specific platform’s target (e.g. Spotify, but not Youtube, Pandora, etc.)—unless you plan on submitting separate files for each target to each different platform.

there was some good discussion on the topic a couple of weeks ago, here:

viewtopic.php?f=5&t=7519158&hilit=LUFS
I write good music for good people

https://slowrobot.bandcamp.com/

PhillipOrdonez
Posts: 3732
Joined: 20 Oct 2017
Location: Norway
Contact:

09 Sep 2020

Billy wrote:
09 Sep 2020
PhillipOrdonez wrote:
08 Sep 2020
As for the targets, unless you're producing acoustic music do not try to master your music to those targets...
Yeah I'm glad you pointed that out but that does make me wonder why they are listed in the presentation?

What general targets would you suggest?

Mine are as follows

Code: Select all

Balanced Mix
INT - Integrated LUFS   -13
 ST - Short Term LUFS   -11

dBTP - True Peak        -3 dBTP 
  dB - Peak             -4 dB

DR - Dynamic Range      10 DR
LU - Loudness            3 LU
Those targets mean nothing, they are only for you to know what each service does in terms of loudness normalisation. Targets for music streaming are not goals at all for musicians. However they are very important for engineers who work in broadcast cause they have to ensure their material hits those targets, otherwise the material will get rejected.

Make your music sound as good as possible, and if it is ever meant to be played at clubs, as loud as possible.

I personally find a healthy level at around -9 lufs, anything near that is fine +/- a couple of dB.

User avatar
Billy+
Posts: 4157
Joined: 09 Dec 2016

09 Sep 2020

I did read through that thread and took a copy of the Mastered Level chart
IMG_2155.PNG
IMG_2155.PNG (243.13 KiB) Viewed 1476 times
My Master targets

Code: Select all

CD-Masters 
INT - Integrated LUFS   -8
 ST - Short Term LUFS   -6

dBTP - True Peak        0 dBTP 
  dB - Peak             0 dB

DR - Dynamic Range      7 DR
LU - Loudness           3 LU
But Ike what's been said there targets not absolutes, has anyone tried to see if it's better to be turned up or down?

PhillipOrdonez
Posts: 3732
Joined: 20 Oct 2017
Location: Norway
Contact:

09 Sep 2020

Billy wrote:
09 Sep 2020
I did read through that thread and took a copy of the Mastered Level chart
IMG_2155.PNG

My Master targets

Code: Select all

CD-Masters 
INT - Integrated LUFS   -8
 ST - Short Term LUFS   -6

dBTP - True Peak        0 dBTP 
  dB - Peak             0 dB

DR - Dynamic Range      7 DR
LU - Loudness           3 LU
But Ike what's been said there targets not absolutes, has anyone tried to see if it's better to be turned up or down?
Much better to be turned down. If they turn it up, they will apply limiting and you don't want that. I'd they don't apply limiting, the result will clip, and your don't want that either.

User avatar
Billy+
Posts: 4157
Joined: 09 Dec 2016

09 Sep 2020

What about platforms that convert to lower bit rates, doesn't that have the potential to cause +0dB peaks.

I.e. Going for cd quality wav files to 192kb mp3

User avatar
Billy+
Posts: 4157
Joined: 09 Dec 2016

09 Sep 2020

Nice find in the other thread.
Side note: Crest Factor = the difference between Peak and VU/RMS levels.

What is nice about Reason is that when the Big Meter is set to VU + Peak and your limiter is set to between -1 and 0 dBFS, the VU portion of the meter automatically "reads" crest factor. So all I do when adding my final mastering limiter is to make sure the VU hits around -12 dBFS on average during the loudest parts. It's so simple, doesn't require any additional metering, and for me "just works".

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

09 Sep 2020

as long as you aren’t clipping to begin with, and leave enough headroom for any further processing that will be applied (I’ve heard Spotify’s transcoding can cause clipping if you deliver files that are hotter than -2db peak, for example). I imagine that’s a pretty safe range to deliver to just about any platform.
I write good music for good people

https://slowrobot.bandcamp.com/

PhillipOrdonez
Posts: 3732
Joined: 20 Oct 2017
Location: Norway
Contact:

09 Sep 2020

Billy wrote:
09 Sep 2020
What about platforms that convert to lower bit rates, doesn't that have the potential to cause +0dB peaks.

I.e. Going for cd quality wav files to 192kb mp3
They all do. -1db true peak should be fine. Test yourself how much is added with each bit rate. The lower the bit rate, the higher the peak increase.

User avatar
Billy+
Posts: 4157
Joined: 09 Dec 2016

09 Sep 2020

So when all is said and done my Balanced Mix Targets should be my Masted Targets and I should probably export as lower quality MP3 rather than CD wav if I'm going to upload to any platforms?

Code: Select all

Balanced Mix
INT - Integrated LUFS   -13
 ST - Short Term LUFS   -11

dBTP - True Peak        -3 dBTP 
  dB - Peak             -4 dB

DR - Dynamic Range      10 DR
LU - Loudness            3 LU
-3dBTruePeak should then give the headroom to nullify the possibility of creating +0dB peaks

PhillipOrdonez
Posts: 3732
Joined: 20 Oct 2017
Location: Norway
Contact:

09 Sep 2020

Billy wrote:
09 Sep 2020
So when all is said and done my Balanced Mix Targets should be my Masted Targets and I should probably export as lower quality MP3 rather than CD wav if I'm going to upload to any platforms?

Code: Select all

Balanced Mix
INT - Integrated LUFS   -13
 ST - Short Term LUFS   -11

dBTP - True Peak        -3 dBTP 
  dB - Peak             -4 dB

DR - Dynamic Range      10 DR
LU - Loudness            3 LU
-3dBTruePeak should then give the headroom to nullify the possibility of creating +0dB peaks
I would not upload a lower quality MP3, no, they will convert it to something else, makes no sense. Wave and let them use whatever codec they use.

User avatar
Billy+
Posts: 4157
Joined: 09 Dec 2016

09 Sep 2020

PhillipOrdonez wrote:
09 Sep 2020
I would not upload a lower quality MP3, no, they will convert it to something else, makes no sense. Wave and let them use whatever codec they use.
Ok I guess that makes sense as you don't know what bit rate your going to end up at.

So does Dynamic Range have any effect on the normalisation i.e. My target is between 10 to 7, if I pushed that to say 14 to 12 would I end up quieter after conversion?

User avatar
mcatalao
Competition Winner
Posts: 1824
Joined: 17 Jan 2015

15 Sep 2020

My interpretation of these tables is that the streaming companies will normalize to those standards, so if you're near that ballpoint, whatever they do won't be as aggressive.

I've pointed it out to the spotify standards (-15 to -13 LUFS) and never had a turn down either from clients or Distrokid (which is the service i use for my independent distros).

User avatar
mcatalao
Competition Winner
Posts: 1824
Joined: 17 Jan 2015

15 Sep 2020

Billy wrote:
09 Sep 2020
PhillipOrdonez wrote:
09 Sep 2020
I would not upload a lower quality MP3, no, they will convert it to something else, makes no sense. Wave and let them use whatever codec they use.
Ok I guess that makes sense as you don't know what bit rate your going to end up at.

So does Dynamic Range have any effect on the normalisation i.e. My target is between 10 to 7, if I pushed that to say 14 to 12 would I end up quieter after conversion?
Depends on what they do, but i'd say if you're on 7 lufs, you're already working with a very small dynamic range. In reality, I think it works backwards. You have a small dynamic range, you have a too high loudness, so they get it down. IF it is direct, you end up with something that may peak at lower values and will have a standard lufs and a very low dynamic range, thus a quieter result.

One thing I'm certain. There isn't a way to get more Dynamic range when you already don't have it (or at least a way that it will sound right). Dynamic range can be specific to genre, but it's how i was saying in my other post. Even if you don't look at these tables as definitive, the nearer you are, less they have to lower it when applying normalization.

Post Reply
  • Information
  • Who is online

    Users browsing this forum: No registered users and 21 guests