The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Limiters and Bass Intermodulation Distortion
Old 10th February 2009
  #1
Gear Maniac
 
fader8's Avatar
 

Limiters and Bass Intermodulation Distortion

Morning!

I recently purchased the Kjaerhus limiter and was noticing an audible difference immediately from some of the other limiters I have been using while experimenting with limiting bass.

I wrote up a new page on my blog that just looks at bass intermodulation with various software limiters and I wanted to get some feedback from y'all on what might be a good "next test" to do before the demo period runs out in a couple of weeks on a couple of these.

Thanks for any suggestions. Also, if there's a better setting for the Voxengo Elephant to get a little better performance, I'm open to ideas to try. Didn't have time to study the manual thoroughly on that one. Some of you may be more experienced with it.

Thanks! Gotta run off, but I'll be back later.
fader8

Edit: A page 2 has been added HERE.

Last edited by fader8; 12th February 2009 at 02:05 PM.. Reason: Add link.
Old 10th February 2009
  #2
Lives for gear
 
macc's Avatar
 

Verified Member
Brilliant - I'll have a think about this!
Old 10th February 2009
  #3
kjg
Lives for gear
 

Quote:
Originally Posted by fader8 View Post
Morning!

I recently purchased the Kjaerhus limiter and was noticing an audible difference immediately from some of the other limiters I have been using while experimenting with limiting bass.

I wrote up a new page on my blog that just looks at bass intermodulation with various software limiters and I wanted to get some feedback from y'all on what might be a good "next test" to do before the demo period runs out in a couple of weeks on a couple of these.

Thanks for any suggestions. Also, if there's a better setting for the Voxengo Elephant to get a little better performance, I'm open to ideas to try. Didn't have time to study the manual thoroughly on that one. Some of you may be more experienced with it.

Thanks! Gotta run off, but I'll be back later.
fader8
Hey great stuff fader8, thank you for all the tests.

Interesting results for the Kjaerhus. Matches with my experiences. Clean and very transparent when not pushed too much and maybe using a little smoothing. Not the one to use for loudness, but great for transparent or smooth taming of a spike here and there.
I'll try to come up with some useful things to test for as well.

regards,
Klaas-Jan
Old 10th February 2009
  #4
Gear Addict
 

Hey fader8, is the elephant the one in blue in that graph versus kjaerhus? If so, that doesnt look so hot! damn

EDIT: question? That value of 1.00 on the elepant. Is that 1ms or what? What is that value exactly. If it is just 1, wouldnt that be way to fast for bass freq's?
Old 10th February 2009
  #5
Lives for gear
 
macc's Avatar
 

Verified Member
The release time on the Elephant is 1ms - not 1 second
Old 10th February 2009
  #6
Gear Addict
 

Quote:
Originally Posted by macc View Post
The release time on the Elephant is 1ms - not 1 second
hi macc, so sorry if i'm mistaken here, is not 1ms way too fast for bass freqs?
Old 10th February 2009
  #7
jdg
Lives for gear
 
jdg's Avatar
 

Verified Member
yah, this is the sound i hate.
try an L2. (dont own one anymore)
Old 10th February 2009
  #8
Lives for gear
 
macc's Avatar
 

Verified Member
Quote:
Originally Posted by somsto75 View Post
hi macc, so sorry if i'm mistaken here, is not 1ms way too fast for bass freqs?
I posted that before your edit Yes it is too fast for bass.

I mentioned it as the test conditions stated that release times were set to 1 second, when the elephant was set to 1ms. So the test is not correct with regard to Elephant, hence it looking… a bit crap heh
Old 10th February 2009
  #9
Gear Addict
 

Quote:
Originally Posted by macc View Post
I posted that before your edit Yes it is too fast for bass.

I mentioned it as the test conditions stated that release times were set to 1 second, when the elephant was set to 1ms. So the test is not correct with regard to Elephant, hence it looking… a bit crap heh
cool thanks, i'm not such a dumbass afterall thumbsup
Old 10th February 2009
  #10
Lives for gear
 
Alexey Lukin's Avatar
 

Verified Member
I'd like to see the resulting WAV files as well. When testing limiters, it's important to set them to equal aggressiveness, i.e. to match output RMS precisely. Otherwise you can easily change levels of intermodulation of any limiter by adjusting its release time or other controls.
Old 10th February 2009
  #11
kjg
Lives for gear
 

Yes, the Elephant needs to be redone or taken out.
Old 10th February 2009
  #12
Lives for gear
 
Alexey Lukin's Avatar
 

Verified Member
Typical release times of mastering limiters are between 1 and 10 ms.
1000 ms seems way too much for a reasonable test.
Also, setting the release time to the same value does not guarantee equal aggressiveness, see my (old and outdated) article.
Old 10th February 2009
  #13
kjg
Lives for gear
 

I thought they were all 1ms at first...

That would be my first suggestion then. To redo the test for all at 10ms release time. 10ms is on the conservative side of realistic I'd say for aggressive limiting settings.
Old 10th February 2009
  #14
Lives for gear
 
Nordenstam's Avatar
 

Verified Member
Nice test!

On the elephant 2, "shape" parameter sets the level of distortion. (as observed using VST plug in analyzer). Haven't tried V3 but I guess it's pretty much the same. Perhaps try to tweak the parameters a bit? I think the elephant can be made to be much softer than default, perhaps even match the Kjærhus in distortion level.
Old 10th February 2009
  #15
Lives for gear
 

Why on earth do you think Elephant v3 is the same as v2? It's like comparing iZotope Ozone version 2.0 to the current 4.0.

Come on.. Voxengo Elephant v3.0 is completely rewritten from scratch and has a VASTLY superior limiting algorithm.

For some presets for version 3, try these:

Voxengo Elephant v3 presets


EDIT: I know this is pointed out at the blog but it's worth mentioning again:

For mastering, the "sinewave through limiter to see distortion" test is flawed, at least in my opinion. Distortion is NOT the worst that can happen to your audio when it is being limited. What happens to the transients is much more important, at least in any popular music genre. If you want to use it as a bass instrument limiter, then perhaps, but otherwise transient response is much more important (and the main weakness of the otherwise excellent Kjaerhhus limiter).

EDIT2: In an ideal world, without the loudness war, these test would probably mean more as we wouldn't be pushing the limiters so hard.

- bManic
Old 10th February 2009
  #16
I second the suggestion of testing the limiters at 10ms -- more of a real-world value, perhaps.

This reminds me of the dreaded "can you add more bass AND make it louder" mastering requests. You know things are getting silly when the limiter's clamping down on the sub-bass... Never mind the transients -- they left town last week!
Old 10th February 2009
  #17
kjg
Lives for gear
 

I choose Elephant when I want the loud and snappy at the price of distortion type of sound. I choose Kjaerhus for relatively smooth and or transparent with just 3 parameters or so.

It is a bit of apples and oranges or different tools for different jobs, although I agree that with tweaking you can probably get the Elephant pretty close to the mpl-1.

Still interesting to compare at equal release times, but somewhat shorter makes more sense IMHO.

regards,
kjg
Old 10th February 2009
  #18
Lives for gear
 
macc's Avatar
 

Verified Member
Quote:
Originally Posted by csl View Post
This reminds me of the dreaded "can you add more bass AND make it louder" mastering requests.
Hahahaha!!

*looks to the sky*

Noooooooooooooooooooooooooooooooo!!!!..........
Old 11th February 2009
  #19
Gear Maniac
 
fader8's Avatar
 

Hey folks, thanks so much for the posts.

Quote:
Originally Posted by Lupo View Post
On the elephant 2, "shape" parameter sets the level of distortion. (as observed using VST plug in analyzer). Haven't tried V3 but I guess it's pretty much the same. Perhaps try to tweak the parameters a bit? I think the elephant can be made to be much softer than default, perhaps even match the Kjærhus in distortion level.
OK, Elephant fixed. I've updated the site. I tweaked it around a lot (and read the manual too) and it's a fascinating limiter. Release is now set to 300mS. The result shown is the best I can do under these steady-state conditions, but it's clear that this thing was built to deal with transients. I still wish it had much better metering.

Quote:
Originally Posted by Alexey Lukin View Post
Typical release times of mastering limiters are between 1 and 10 ms.
1000 ms seems way too much for a reasonable test.
Agreed in principle, but not all mastering work is created equal. Much of the work I get is more like restoration & mastering, and much of it isn't even music. I'll often chop something into bands and have to work on them in isolation. There are actually times when those long release times are exactly what I need. For some of the sound design work I do, it's absolutely invaluable.

Quote:
Originally Posted by kjg View Post
That would be my first suggestion then. To redo the test for all at 10ms release time. 10ms is on the conservative side of realistic I'd say for aggressive limiting settings.
That's a good idea. I'll plan another page for that. Now that everything is "presetted" it's cake to redo with new settings.

I'll stress again that these tests may not individually reflect real-world mastering conditions, but only provide insight into the process behaviour under specific conditions. All the more reason to look deeper and try different things. I don't think any of these products are particularly bad or good, but they likely all have particular application where they really shine, and that's the goal, to find the shiny bits!


Alexey, I'm not sure I grok your use of the term "aggressive". In a steady state limited signal, the only aggression I can see is simply the amount of gain reduction, it's waveshape as a result of attack time and look ahead perhaps, but I'm sure I'm missing something. What's confusing me a little is that I can see dramatic change in the intermod products with this test signal when I change the release time from say, 100mS to 300mS. My gut tells me that with this signal, with no dynamic changes present at the detection algorithm, I should see no difference in the limiters operation. But I do. So thus, I am missing something fundamental here.

Thanks again, guys.
Old 11th February 2009
  #20
Lives for gear
 
Alexey Lukin's Avatar
 

Verified Member
Aggressiveness means the amount of loudness that limiter achieves at the specified threshold level. You are trying to match limiters by approximate peak levels and release times (which are very manufacturer-variable) in your test. My suggestion is measuring RMS instead, and matching RMS precisely, down to 0.01 dB, only by means of release/shape/character/pdr controls, while the threshold is the same for all limiters. It is pretty obvious that limiters with higher RMS will also generate more distortion, but the spectral shape of this distortion is the matter of interest.

For example, the fact that Elephant produced more distortion than Kjaerhus in your test probably only means that Elephant has been pushed harder (working in a more aggressive mode, producing higher RMS) than Kjaerhus.
Old 11th February 2009
  #21
Gear Maniac
 
fader8's Avatar
 

Quote:
Originally Posted by Alexey Lukin View Post
Aggressiveness means the amount of loudness that limiter achieves at the specified threshold level. You are trying to match limiters by approximate peak levels and release times (which are very manufacturer-variable) in your test. My suggestion is measuring RMS instead, and matching RMS precisely, down to 0.01 dB, only by means of release/shape/character/pdr controls, while the threshold is the same for all limiters. It is pretty obvious that limiters with higher RMS will also generate more distortion, but the spectral shape of this distortion is the matter of interest.
OK, I'm looking at it the other way, I guess. I'm assuming all these limiters are reducing the crest factor of the sine wave's fundamental the same. Because the PPM is making a broadband measurement, it's seeing the intermodulation distortion as additional signal content that's adding to the measured rms value.

Quote:
Originally Posted by Alexey Lukin View Post
For example, the fact that Elephant produced more distortion than Kjaerhus in your test probably only means that Elephant has been pushed harder (working in a more aggressive mode, producing higher RMS) than Kjaerhus.
Ah, I'm getting your language now. You're equating an increase in loudness to an increase in rms value for a given peak value. You're referring to aggression as a decrease in crest factor, regardless of its peak value. I think I'm on your page now.

So, that leaves us with what seems to be a nebulous term then, "Gain Reduction", which is apparently not so well defined. Or at least, not implied to mean the same process that's occurring. Because in all these cases, exactly 1dB of gain reduction is occurring. But this would only refer to its peak value. If I'm following your logic: Limiter A and Limiter B both provide 1dB of gain reduction, but A results in XdB of crest factor reduction and B results in YdB of crest factor reduction.

Let me try a few things and I'll get back.
Old 11th February 2009
  #22
Mastering
 

Quote:
Originally Posted by fader8 View Post
is occurring. But this would only refer to its peak value. If I'm following your logic: Limiter A and Limiter B both provide 1dB of gain reduction, but A results in XdB of crest factor reduction and B results in YdB of crest factor reduction.

Let me try a few things and I'll get back.
It's largely a matter of distortion and (minutely) increased distortion does mean less crest factor. The faster the release time the more distortion, especially in the bass end.

The L2 has an auto release time which has been optimized more in the direction of increased apparent loudness by tolerating a faster release time more often than a slower one, compared to some of the competition. The TC brickwall limiter has a number of auto release time algorithms. Guess which one sounds like the L2 and purposely has a faster (on the average) more aggressive release time and more distortion? The algorithm called "loud". Yes, it was all on purpose.

It's hard to notice these differences at 1 dB of gain reduction, but the more you push it, the more that the differences you perceive are related to amount of distortion. At least with the limiters that react within a sample and have some form of auto release time. When you switch the release time to manual (if this is available) you can reduce the distortion, but the more you push the limiter, with this slower release time, you're liable to hear clamping effects---where the limiter doesn't recover fast enough.

Those are the tradeoffs. Unless everybody backs off (fat chance).

BK
Old 11th February 2009
  #23
kjg
Lives for gear
 

Dear fader8,

Let me first repeat that I really appreciate that you are doing these tests.

Here is a thought; I am not sure about it: Maybe you should aim for a bit more gain reduction to exaggerate the differences slightly? To push the limiters a bit more?

2dB would be too much maybe, but 1.4 or 1.6?

1.6 dB at 10 ms release seems to me like more realistic real world "limiting for loudness" use, then 1 dB at 1000.

Btw: Although I agree that ideally we shouldn't use limiters for loudness, the real life situation is that almost all of us have to sometimes. We might as well be informed on the different options of doing it, so we can choose the lesser evil for the track in question.
Also, as mentioned in other threads, the character of a brickwall limiter and its artifacts has become part of the sound of some music styles.

regards,
Klaas-Jan
Old 11th February 2009
  #24
Gear Maniac
 
fader8's Avatar
 

Quote:
Originally Posted by kjg View Post
Dear fader8,

Let me first repeat that I really appreciate that you are doing these tests.

Here is a thought; I am not sure about it: Maybe you should aim for a bit more gain reduction to exaggerate the differences slightly? To push the limiters a bit more?

2dB would be too much maybe, but 1.4 or 1.6?

1.6 dB at 10 ms release seems to me like more realistic real world "limiting for loudness" use, then 1 dB at 1000.

Btw: Although I agree that ideally we shouldn't use limiters for loudness, the real life situation is that almost all of us have to sometimes. We might as well be informed on the different options of doing it, so we can choose the lesser evil for the track in question.
Also, as mentioned in other threads, the character of a brickwall limiter and its artifacts has become part of the sound of some music styles.
Well said, Klaas. I've been playing with these things a bit, and before I launch into a new battery of tests I want to be sure I'm comparing apples to apples. For example, one of the reasons I picked such a long release time for the initial testing was to try and avoid the differences in how release is initially triggered from one limiter to another. The Oxford Limiter manual clearly states that the release phase is triggered when the detection signal drops 10dB. Apparently, this is not necessarily the case with all of them! Where's that hysteresis control when you need it, eh?

And by the way, having you call me "Dear" just makes all of this worth it!
Old 11th February 2009
  #25
Gear Maniac
 
fader8's Avatar
 

Quote:
Originally Posted by bob katz View Post
It's largely a matter of distortion and (minutely) increased distortion does mean less crest factor.
Indeed it is the distortion. I just tried the Oxford with 1.6dB GR and 10mS release. I lose .5 dB from the crest factor. After inserting Cambridge 6 pole ellipticals after the limiter to confine the bandpass to 33-240Hz, I got the .5 dB back!
(Recall that my test frequencies are 40 and 185Hz.)

Quote:
Originally Posted by bob katz View Post
The faster the release time the more distortion, especially in the bass end.
OK Bob, I think I'm getting clearer on the micro-operation of these things. Knowing that the limiter retriggers its attack phase for every single wave crest that passes its threshold, It hadn't occurred to me that conversely, it initiates its release phase when each crest drops to its -10dBr value, (in the Oxfords case). I had thought that it might wait a little and measure again, but it's obvious to me now. I can see the value of the "Hold" feature on the Kjaerhus limiter. At least in certain situations.

Quote:
Originally Posted by bob katz View Post
It's hard to notice these differences at 1 dB of gain reduction, but the more you push it, the more that the differences you perceive are related to amount of distortion. At least with the limiters that react within a sample and have some form of auto release time. When you switch the release time to manual (if this is available) you can reduce the distortion, but the more you push the limiter, with this slower release time, you're liable to hear clamping effects---where the limiter doesn't recover fast enough.
I got ya. So my question now is, are the ARC and PDR features really more like program dependent "hold" functions? In other words, anticipating new peaks and inhibiting a release phase?
Old 11th February 2009
  #26
Mastering
 

Quote:
Originally Posted by fader8 View Post

I got ya. So my question now is, are the ARC and PDR features really more like program dependent "hold" functions? In other words, anticipating new peaks and inhibiting a release phase?
You are very perceptive! To be honest, I've not spent much time or energy on optimizing PDR and hold parameters on my limiters. I try to keep my limiting to 1 dB or so, and so far the TC in "loud" lets me do that. I suppose I lose a few shootouts that way but that's life. In my copious free time I'm going to investigate the Xenon to see what more I can get out of it. I haven't even learned how to use its second stage yet. Get ready for blastoff! And there is no fancy hold circuitry in there at all.

Usually release hold circuitries are for compressors, not limiters. With limiters you want them to attack fast and release fast, get in there, be invisible and get out of there fast. Otherwise you start punching holes in your music. Can't beat the laws of physics :-).
Old 12th February 2009
  #27
Gear Maniac
 
fader8's Avatar
 

OK, by popular demand, page 2 is HERE and includes a few more limiters.

10mS release times, 1.6dB gain reductions.
Old 12th February 2009
  #28
Lives for gear
 
macc's Avatar
 

Verified Member
Very interesting, thanks again
Old 12th February 2009
  #29
Here for the gear
 
domestic705's Avatar
 

Quote:
Originally Posted by bmanic View Post

For mastering, the "sinewave through limiter to see distortion" test is flawed, at least in my opinion. Distortion is NOT the worst that can happen to your audio when it is being limited. What happens to the transients is much more important, at least in any popular music genre. If you want to use it as a bass instrument limiter, then perhaps, but otherwise transient response is much more important (and the main weakness of the otherwise excellent Kjaerhhus limiter).

EDIT2: In an ideal world, without the loudness war, these test would probably mean more as we wouldn't be pushing the limiters so hard.

- bManic
Hey guys !

wouldn't it be a good idea to use 2 limiters or 1 limiter and a clipper for a better and cleaner sound when trying to get an extremely loud masters?

i mean i usually dont use limiters when mastering cuz from my tests all limiters are transient killers compared with clipping ,
but this test got me thinking that using one of the cleaner limiters such as Kjaerhus or the Precision Limiter (not pushed to hard) then clipping OR using a more 'non-transient killer' limiter such as the Oxford may result a very clean and loud master.

what do u guys think?

ps. i know its stupid to fight peaks , but still my clients "want it loud"
Old 12th February 2009
  #30
Lives for gear
 
wado1942's Avatar
 

Well I've decided that I don't want anything over -9dBfs RMS to leave my studio with my name on it. But I DO think using a soft clipper in conjuction with a limiter often gives better results. Use a soft clipper to reduce the sharp, narrow spikes that drive a limiter crazy, then use the limiter to get the master the rest of the way there. Yeah, the distortion's a bit higher but it doesn't sound as wimpy as using a limiter alone.
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
whosyourdaddy00 / So Much Gear, So Little Time
15
ryangeller / Geekslutz Forum
15
Ski spiggy / So Much Gear, So Little Time
2
vicprobey / So Much Gear, So Little Time
14

Forum Jump
Forum Jump