Gearslutz

Gearslutz (https://www.gearslutz.com/board/)
-   Mastering Forum (https://www.gearslutz.com/board/mastering-forum/)
-   -   Limiter comparisons (with sound) (https://www.gearslutz.com/board/mastering-forum/143991-limiter-comparisons-sound.html)

Lagerfeldt 10th September 2007 05:21 PM

Limiter comparisons (with sound)
 
WHAT
Here's a little something I toyed around with earlier today. A limiter comparison.

Thanks to all those who contributed!

The same track at approximately the same level of RMS (+/- a few fractions only).

Posted with permission from the artist/label naturally.

DIY - guide lines
I've included the original 24 bit master without limiter so you can test with your favorite limiter or favorite settings. But aim for the same RMS level - so do your own level check of my files and your file. Output level is -0.12 dBFS or as close as possible. The straight clipping file (E) was not lowered, so that one peaks at 0 dBFS.

If anyone wants to play along please DO NOT post your version here but send me a YouSendit.com or server link (via PM) and I'll check the RMS value and provide a link in this post.

If you feel you can do an overall better version using one of the already included limiters please let me know and I'll swap my version for your better version.

NO other processing in any way other than the limiter is allowed apart from dithering (your choice but POW-r#3 if possible). Remember it's gotta be a 16 bit file.

DOWNLOAD
Included in my session are the following files (for now):

[files removed, check later posts for new links]

A Sonnox Oxford Limiter
B Waves L2
C Waves L3 UltraMaximizer
D Flux Pure Limiter
E Sonalksis FreeG clipping
F Voxengo Elephant 3
G Ozone Limiter
H Timeworks Master Compressor clipping
I UAD Precision Limiter

X Original 24 bit master without limiter

MY OWN OBSERVATIONS
A Lots of nice transients but also a tad pumpy which suits the track fine in this case
B Slightly muddy and significant loss of transients
C Woah, this is not good. Basic Profile with manual release.
D Very nice and less pumping than A but also slightly less transient info remains than in A and G
E Straight clipping works well here but some harshness in the top and less low level detail/less warmth is the price you pay
F Somewhat like L2 but less distortion and over all somewhat better sound
G Very little pumping and nice fullness in the sound, surprisingly much transient info seems to remain
H Too much coloration in the upper mids and somewhat sterile but a bit of clipping is fine
I Slightly too much coloration overall, and a loss of transients but otherwise fine

I'd probably go for A (Sonnox Oxford) or G (Ozone Limiter). If the client wanted even hotter I'd add 0.5 to 1 dB of clipping post fader. In some cases the Flux Pure Limiter (D) would be my choice instead.

Cheers!

Ditherman 10th September 2007 07:24 PM

A through E all pretty much sound the same to my ears.

Lagerfeldt 10th September 2007 08:28 PM

This requires a bit of practice and good monitoring of course.

Try first listening to one file at a time from start to finish.

Then try aligning them all in a DAW and switch between them listening for transient response (peaks, attack times, notice that loss of transients leads to a slightly dull high end, and clipped transients can lead to harsh top end).

Listen for overall type and quality of sound, distortion (usually caused by release times), stereo perspective (caused by linked or unlinked L/R and other factors), volume vs. punch, etc.

A good trick is to first focus on overall sound, then focus on individual elements (such as the click in the kick drum) and then go back in focus to the overall picture. All the time switching back and forth between versions.

Hope that helps.

ascottk 10th September 2007 08:35 PM

Quote:

Originally Posted by Lagerfeldt (Post 1488505)
WHAT

A Sonnox Oxford Limiter
B Waves L2
C Waves L3 UltraMaximizer
D Flux Pure Limiter
E Sonalksis FreeG clipping
F Original 24 bit master without limiter

MY OWN OBSERVATIONS
My own comments so far are:

A Lots of nice transients but also a bit of pumping which I think suits the track fine
B Slightly muddy and clearly loss of transients
C Woah, bad. In all fairness I used the Basic Profile.
D Very nice and less pumping than A but also slightly less transient info than A
E Straight DAW clipping works surprisingly well here but some harshness in the top

Flux is clearly a worthy candidate for the limiter crown IMO

Cheers!

A Seems to retain more dynamics, nice & clear, & sounds fuller
B Agreed
C This is the muddiest of the bunch
D Agreed
E Agreed

I tried voxengo elephant although I don't think the RMS levels are correct. I'll send you the file. My impressions are that elephant has some aliasing going on, but there's a lot of different settings to play around with so that might be remedied.

Other than the aliasing, it's not as full as the Sonnox but more detailed than the Waves. Elephant seems to retain most of the dynamic content (which it was designed to do). I'll probably experiment more to try & decrease aliasing & retain more of the high freqs.

The settings I used in elephant are:
  • In = +2.0 dB
  • Out = -0.3dB
  • Shape = 0.00
  • RShape = 20 ms
  • Lim Mode = El-3
  • Lim Speed = Med
  • Dithering = OFf
  • DC Filter = Off

More info here:
http://www.voxengo.com/doc/elephant/

Lagerfeldt 10th September 2007 09:05 PM

I haven't checked the file you sent yet but I think you should set it for -0.12 dBFS not -0.3 dBFS and adjust in/threshold to get the right RMS or it won't be a match in this particular test.

More files are welcome.

squeegybug 10th September 2007 09:07 PM

Why not blind testing?

Steve

Lagerfeldt 10th September 2007 09:17 PM

Let's keep things in the open, it's better for discussion.

If somebody wants to blind test, it's possible on their own or if they have a friend helping.

I think the differences are pretty obvious and blind testing doesn't make much sense in this test IMO.

UnderTow 10th September 2007 10:44 PM

Quote:

Originally Posted by ascottk (Post 1488903)
The settings I used in elephant are:
  • In = +2.0 dB
  • Out = -0.3dB
  • Shape = 0.00
  • RShape = 20 ms
  • Lim Mode = El-3
  • Lim Speed = Med
  • Dithering = OFf
  • DC Filter = Off

These settings won't match the RMS levels of the other files. I'm uploading a file for lagerfeldt as we speak. (My version also has more aggressive settings which I think are more appropriate for this style).

Alistair

ascottk 10th September 2007 11:02 PM

Quote:

Originally Posted by Lagerfeldt (Post 1488977)
I haven't checked the file you sent yet but I think you should set it for -0.12 dBFS not -0.3 dBFS and adjust in/threshold to get the right RMS or it won't be a match in this particular test.

More files are welcome.

Quote:

Originally Posted by UnderTow (Post 1489159)
[/LIST]These settings won't match the RMS levels of the other files. I'm uploading a file for lagerfeldt as we speak. (My version also has more aggressive settings which I think are more appropriate for this style).

Alistair

Okay, I misread the rules gooof

dynomike 11th September 2007 04:44 PM

The Oxford (A) is the clear winner to my ears. The clipper (E) actually seems to be the next best thing in this case...

To be fair, the L3 (ultra) does sound a lot better when you play with the release time and shaping controls to better match the program. I'll have to look further into the oxford plugs though. Thanks for the comparison!

Mike

Lagerfeldt 11th September 2007 07:28 PM

I only used the L3 UltraMaximizer but I did adjust release time to fit the song. All of the presets (only user adjustable in the L3 MultiMaximizer) sounds pretty bad.

Coming up are Voxengo Elephant and Timeworks tomorrow.

Also I'm changing the Flux Pure Limiter file with an identical copy but with 1.45 ms look ahead (0 look ahead in the first file).

Luna Sound 12th September 2007 09:44 AM

I've had a quick listen and I definitely like C (L3) the least. I also agree that A (Sonnox limiter) sounds the best so far. On this material E (clipping) seems like a really viable option. I still prefer the sound of A but only just!

Tubefreak 12th September 2007 12:05 PM

Great stuff, have to check it in my studio to see which I like best. Unfortunatly I won't be in the studio for a few days.

The UAD precision limiter sounds very nice imo and would be great to add that to this comparison. But it will be a few days before I'm back in the studio to do make a file.

Anyone who else owns the UAD Prec Lim who's able to make a track?

Kind regards,
Maarten

Lagerfeldt 12th September 2007 01:49 PM

I have a Voxengo Elephant version for download later tonight.

I still need a Sonic Timeworks Mastering Compressor version without clipping going on. Anyone?

UAD Precision Limiter too. Please read the rules and send me a link.

Luna Sound 12th September 2007 02:18 PM

I've pm'd you.

MattGray 12th September 2007 03:27 PM

I've just done an Ozone3 version for you which I'm uploading to YouSendIt now. I'll PM you the link shortly. One thing I noticed that on analyzing the RMS levels of your samples there is a deviation of up to 0.3db RMS which is enough to be noticeable, for instance the L3 sample is -9.6db RMS where as the Oxford is -9.3db RMS. Just thought I'd mention it...

Matt

scraggs 12th September 2007 07:00 PM

Quote:

Originally Posted by Lagerfeldt (Post 1492328)
I still need a Sonic Timeworks Mastering Compressor version without clipping going on. Anyone?

isn't the timeworks essentially just a clipper? tracks were always flat-topped when i used it. still liked it way better than l2 though...

UnderTow 12th September 2007 09:22 PM

Quote:

Originally Posted by MattGray (Post 1492553)
I've just done an Ozone3 version for you which I'm uploading to YouSendIt now. I'll PM you the link shortly.

Heh, I sent him one too. :) What settings did you use? I used:

Character: Very Fast (0.5)
Prevent inter-sample clips enabled
Mode: Intelligent
Threshold: -4.7
Margin: -0.1
DC and dithering off. (I use POW-R 3)

Quote:

One thing I noticed that on analyzing the RMS levels of your samples there is a deviation of up to 0.3db RMS which is enough to be noticeable, for instance the L3 sample is -9.6db RMS where as the Oxford is -9.3db RMS. Just thought I'd mention it...
Indeed but it is very hard to set things up so that they end up absolutely identical. These are the values I have for all the tracks in order of RMS values: (Some not published yet).

Peak RMS L RMS R

Oxford: -0.11 -6.85 -6.72
Epure: -0.11 -6.9 -6.79
Ozone: -0.24 -6.91 -6.79
L2: -0.09 -6.98 -6.86
Sonalksis: 0 -7.02 -6.89
Elephant 02: -0.16 -7.06 -6.97
L3: -0.09 -7.18 -7.05


Alistair

MattGray 13th September 2007 12:16 AM

Quote:

Originally Posted by UnderTow (Post 1493513)
Heh, I sent him one too. :) What settings did you use? I used:

Character: Very Fast (0.5)
Prevent inter-sample clips enabled
Mode: Intelligent
Threshold: -4.7
Margin: -0.1
DC and dithering off. (I use POW-R 3)

Ha, great minds think alike! These are the settings I used...

Character : Very Fast (0.3)
Prevent inter-sample clips enabled
Mode: Intelligent
Threshold: -4.6
Margin: -0.12
DC and dithering off. (I used the MBIT+ dither on 16bit Ultra/High)

I figured it's part of the Ozone sound the dither, so why not use it.

Matt

dynomike 13th September 2007 07:04 AM

Yeah, you should put up the Timeworks file I sent you. Thats pretty much the cleanest it gets with the Timeworks.. RMS and peak levels are the same as the oxford, so it should be a fair comparison. I wasn't *trying* to clip it, thats just what happens with that plugin. Not a lot of shaping going on there..

Interested to hear the precision limiter, etc! I have a UAD-1, it'd be great to know if that plugin is gonna improve things substantially for me.

Mike

Tubefreak 13th September 2007 07:31 AM

In the past I've used the Timeworks Mastering Compressor (= limiter). But I get (far) better results with the UAD Precision limiter, that one really can take a beating before sounding awful.

Although it did sound bad at all, I really disliked the fact that you can't see what your settings are. I want to know exactly how much the gain reduction, output level, etc are set at and the TMC did not allow me to do that.


A UAD Precision Limiter file has been added.

UAD-1 Precision Limiter
Input 4.95 dB
Output -0.12 dB
Release Auto
Mode A (better transients then B imo)

Did no DC Offset removal, since the original files all had DC too.

Psycho Dither
16 bit
noise shape on
dither on (high pass TPDF)
dither bit width 2.0

Maarten

Stiff 13th September 2007 07:42 AM

Great post, very useful for me at the moment since I'll be reviewing Flux Pure Limiter soon and the ones I've compared it to so far has been plug-ins you haven't mentioned here (Massey, McDSP, Digidesign). kfhkh

Bob Yordan 13th September 2007 07:59 AM

thumbsup Cool thread, will listen to the files tonight and see if I can make some contribution?

Did just have a short peak at the flux & Sonnox files via $6 head phones on my work PC.
It seems like the Flux boosts the high mid a bit more than the Sonnox. Some 'effect' sounds
that appear from time to time gets a bit clearer, but at the same time the bass is a tiny
bit thinner. Will be interesting to hear tonight what the b&w speakers will reveal?

Would be really cool if someone could run it via the TC Powercore mastering brickwall limiter
(system 6000 based ) also.

Best wishes

Lagerfeldt 13th September 2007 09:24 AM

Coming up later tonight are:

UAD Precision Limiter
Ozone Limiter
Voxengo Elephant
Timeworks

I'd prefer not to include the Timeworks as that's more like a clipper but I guess I will. I only included the Sonalksis FreeG clipper as a reference, not as an alternative to limiting in this test.

I.e. if you want clipping you could still use the e.g. Sonnox Oxford limiter + clipping which might yield a better result than just clipping. Or not. :-)

sedohr 13th September 2007 10:22 AM

Looking forward to see how the Massey compares..Anyone ?

Kalli

Lagerfeldt 13th September 2007 07:09 PM

Okay, I've changed the download so the files are available on an individual basis.

If you already downloaded be sure to download file D again and replace, as look ahead time has been fixed now.

Please look in the original post above for the links.

I was very surprised by the Ozone Limiter (as I tested it myself some time ago but wasn't impressed). At least in this test it fits the job very well.

I received several files for both the Ozone and UAD, and I picked the best sounding ones.

So far I prefer (rated)

G Ozone Limiter
A Oxford Limiter
D Flux Pure Limiter
I UAD Precision Limiter
F Voxengo Elephant
B Waves L2
C Waves L3 UltraMaximizer

I'm not rating the two clipped files:
E Sonalksis Free G
H Sonic Timeworks Master Compressor

I would prefer to use limiting such as in G or A and then clipping in the DAW if I wanted clipping too, but that's just me perhaps.

aivoryuk 13th September 2007 08:29 PM

Quote:

Originally Posted by Lagerfeldt (Post 1495670)
I would prefer to use limiting such as in G or A and then clipping in the DAW if I wanted clipping too, but that's just me perhaps.

well that is kind of what you're getting with example G a mixture of clipping and limiting (more clipping though)

0.0 (character) on the ozone is clipping and moving the character slider up brings it more into a limting mode. It has a bricklimiting mode as well. Matt's setting of 0.3 suggests its acting more of a clipper then a pure limiter

Obv you outlined people could use whatever settings they wanted but from a objective point of view it may be worth just asking for a sample of the ozone just in straight limiting mode.

Lagerfeldt 13th September 2007 08:36 PM

Right, good point.

A nice feature in the Ozone then.

That would explain why I never got similar results when I used it, obviously I only went for pure limiting.

What would lower my opinion of the Ozone regarding pure limiting. However, looking at the waveform is there really that much clipping going on overall?

This is definitely a grey area, and I would urge people to draw their own conclusions.

aivoryuk 13th September 2007 08:51 PM

Quote:

Originally Posted by Lagerfeldt (Post 1495919)
Right, good point.

A nice feature in the Ozone then.

That would explain why I never got similar results when I used it, obviously I only went for pure limiting.

What would lower my opinion of the Ozone regarding pure limiting. However, looking at the waveform is there really that much clipping going on overall?

This is definitely a grey area, and I would urge people to draw their own conclusions.

i don't think the ozone in pure limting would lower your opinion at all, overall the loudness maximizer in the ozone is extremley good. I had never considered really it (mainly due to hype that a lot of people consider ozone to be a joke) until Matt Gray and other people were saying good things about it. I tried it i liked.

(off topic i'm quite eager to check out Izotopes SRC as on paper it looks great, but don't where its available)

My main point was that making a subjective opinion from something that just does limiting like the L2 compared to the ozone that has a lot more options for choice is quite hard as they are 2 different beasts.

although i've not check out the sonnox limiter properties i may have to check that one out.

fwiw i don't have the flux limiter but have the solera with the clipping mode switched on whcih i think sounds great

Bob Yordan 13th September 2007 09:54 PM

Quote:

Originally Posted by Tubefreak (Post 1494627)
In the past I've used the Timeworks Mastering Compressor (= limiter). But I get (far) better results with the UAD Precision limiter, that one really can take a beating before sounding awful.

Although it did sound bad at all, I really disliked the fact that you can't see what your settings are. I want to know exactly how much the gain reduction, output level, etc are set at and the TMC did not allow me to do that.


A UAD Precision Limiter file has been added.

UAD-1 Precision Limiter
Input 4.95 dB
Output -0.12 dB
Release Auto
Mode A (better transients then B imo)

Did no DC Offset removal, since the original files all had DC too.

Psycho Dither
16 bit
noise shape on
dither on (high pass TPDF)
dither bit width 2.0

Maarten

Hiya

I tried the UAD precision limiter inserted after one of my own developed plugs and did hit it with 6 to 7dB and it sounded rather clean and only made one major clipp. The RMS average finished at -7.7dB, processing the not processed wav. Min was -59dB and max was -3.5dB and peak -0.12 dB according to WL6.

One of my many silly plug experiments. gooof heh

howdy