R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: [1]   Go Down

Author Topic: Clipping & oversampling modes, audible differences  (Read 8299 times)

Viitalahde

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 99
  • Real Full Name: Jaakko Viitalähde
Clipping & oversampling modes, audible differences
« on: April 01, 2011, 01:54:25 pm »

Now this has been puzzling me for quite a while, and while I'm not sure any universal conclusions can be made since I'm talking of what one software does, I'm still interested in hearing any input.

For digital clipping, I use the Voxengo Elephant. Sometimes you need to clip pretty hard, and I find it fascinating how different oversampling modes give slightly different tones. The stereo image might become wider or the sides more separated, the harmonic structure of distorted instruments changes slightly, things like that.

What's even stranger is that bigger isn't always better. I might find that 2X doesn't differ much from 1X, 3X is noticably wider, 4X gets weird sounding in the highs.. Sometimes no oversampling sounds best, sometimes 8X sounds the best.

I'm pretty sure the answer is in the low pass when going back to 1X, but still.. Is there any basic theory that can explain this?
Logged
Jaakko Viitalähde
Virtalähde Mastering, Kuhmoinen/Finland
http://www.virtalahde.com
http://www.facebook.com/VirtalahdeMastering

SafeandSoundMastering

  • Full Member
  • ***
  • Offline Offline
  • Posts: 222
  • Real Full Name: Barry Gardner
  • My analogue rack
Re: Clipping & oversampling modes, audible differences
« Reply #1 on: April 01, 2011, 02:13:35 pm »

When I clip ITB the first thing I notice with my clipper of choice is an increase in HF.

I rarely clip when not in 2 x oversampling cause I think it sounds better to my ear.

I have Elephant and yet I have never really tried the clipper. One thing about Elephant is it has a very strident, slightly thick, character in the limiting modes, it sounds quite compressor like to me and the OS modes do not always improve the sound. Sure it's flexible but the second I engage it I hear it do something to the music and sadly often it is not something good. I choose it on occasion as it just works for some tracks but it is one of the least transparent limiters I am aware of.

I went through a phase of outright disliking it, hearing sonic degradation the second I put it in place with virtually no GR.

You should try some other clippers IMO, they also sound completely different from each other.

Logged
Barry Gardner
SafeandSound Mastering UK based online mastering studio.

lowland

  • Full Member
  • ***
  • Offline Offline
  • Posts: 110
Re: Clipping & oversampling modes, audible differences
« Reply #2 on: April 02, 2011, 05:24:05 am »

I've said this before, but clipping the TC6000 works here when needed, sounds surprisingly transparent to my ear. I used to clip the dbx Quantum when I had that: could never get on with its hard clip but soft clip was OK for me, the opposite of what Brad found IIRC. Now with the TC hard clipping gives me less obtrusive results than either Quantum mode, though I don't know why that should be (never use the TC's soft clip, BTW): this is necessarily subjective as I no longer have the Quantum.
Logged
Nigel Palmer
Lowland Masters
www.lowlandmasters.com

Waltz Mastering

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 57
  • Real Full Name: Tom Waltz
Re: Clipping & oversampling modes, audible differences
« Reply #3 on: April 02, 2011, 04:41:31 pm »

I don't use Elephant, but on Pro-L there is something that happens when switching to 2x and 4x that makes me very hesitant to use those modes... I'd like to know what's going on under the hood as well.

Cass Anawaty

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 61
Re: Clipping & oversampling modes, audible differences
« Reply #4 on: April 02, 2011, 05:08:38 pm »

I don't use Elephant, but on Pro-L there is something that happens when switching to 2x and 4x that makes me very hesitant to use those modes... I'd like to know what's going on under the hood as well.
I "feel" there's something I don't like about Elephant's oversampling, and I don't use it.
Logged
Cass Anawaty, Mastering Engineer
www.sunbreakmusic.com

Patrik_T

  • Newbie
  • *
  • Offline Offline
  • Posts: 21
Re: Clipping & oversampling modes, audible differences
« Reply #5 on: April 02, 2011, 08:20:46 pm »

.
Logged

SafeandSoundMastering

  • Full Member
  • ***
  • Offline Offline
  • Posts: 222
  • Real Full Name: Barry Gardner
  • My analogue rack
Re: Clipping & oversampling modes, audible differences
« Reply #6 on: April 03, 2011, 09:17:12 am »

It seems some upwards OS modes have been done that subjectively "enhance" and some that
seem to smear and blur. You have to pick and choose.
Logged
Barry Gardner
SafeandSound Mastering UK based online mastering studio.

Herbeck

  • Newbie
  • *
  • Offline Offline
  • Posts: 29
  • Real Full Name: Ulf Herbeck
Re: Clipping & oversampling modes, audible differences
« Reply #7 on: April 03, 2011, 12:33:01 pm »

I've been working on this problem for a while, with both oversampling and SRC.
Here is a small SRC shootout with Saracon.(44,1/24 to 88,2/32 float)

(44,1/24 to 88,2/32)
https://download.yousendit.com/MEtRb241bWc4aU5jR0E9PQ
 
(Saracon 88,2/32)
https://download.yousendit.com/MEtSeFVjR3NCSnBjR0E9PQ

I've been using a similar technique for oversampling with very good results.
I've been able to work with 4x oversampling with very small amounts of artifacts.

Cheers,

Herbeck
Logged

Hermetech Mastering

  • Full Member
  • ***
  • Offline Offline
  • Posts: 181
  • Real Full Name: Gregg Janman
  • Manual Gain Rider
Re: Clipping & oversampling modes, audible differences
« Reply #8 on: April 03, 2011, 07:39:26 pm »

I usually prefer the sound of Elephant at 4x OS to 1x or 2x. Never use 8x though!

Herbeck

  • Newbie
  • *
  • Offline Offline
  • Posts: 29
  • Real Full Name: Ulf Herbeck
Re: Clipping & oversampling modes, audible differences
« Reply #9 on: April 08, 2011, 09:27:01 am »

Voxengo plugins has a bypass knob, however the oversampling is still engaged.
So it's quite simple to hear or to do a null test to see how much mess the oversampling does.


Cheers,

Herbeck
Logged

subvert

  • Newbie
  • *
  • Offline Offline
  • Posts: 15
  • Real Full Name: Ben Williams
Re: Clipping & oversampling modes, audible differences
« Reply #10 on: April 08, 2011, 09:43:08 am »

I sometimes (pending judgement on audition) use oversampling modes on Elephant and Pro-L, when the source is at 44.1 or 48, but not with sources at higher SRs
Logged
Subvert Central Mastering
http://www.scmastering.com

bblackwood

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 254
Re: Clipping & oversampling modes, audible differences
« Reply #11 on: April 11, 2011, 12:33:55 pm »

*temporary bump while thread porting from old forums happens*
Logged
Brad Blackwood
euphonic masters

nisilen

  • Newbie
  • *
  • Offline Offline
  • Posts: 3
  • Real Full Name: Niklas Silen
Re: Clipping & oversampling modes, audible differences
« Reply #12 on: April 11, 2011, 07:22:55 pm »

For digital clipping, I use the Voxengo Elephant.

Keep in mind that Elephant's "clipping" isn't really true clipping. It's a pseudo-soft/limiter hybrid kind of thing. Try feeding it some blocks of a digital constant (a fully positive DC signal, aka a "mega pulse" and look at the results.. there is a release stage in it). It sounds awesome on some material though.

You can get something slightly similar in Pro-L by using either the 'punchy' or 'transparent' modes and setting attack to maximum (10s) and release to zero. It will be almost clipping but still have some program dependent quick release.

The various ovesampling "sounds" is due to the way the linear phase filter reacts to transients. Like all linear sampling processes there will be some pre-ring and depending on the oversampling amount this pre-ring will settle on different parts of the spectrum. Internal parameters will also react differently depending on oversampling. The envelope followers are running at much higher sample rates and because of the complex nature of a modern limiter there are a lot of tiny changes within the plugins.. at least this is my guess when it comes to Elephant.

I've been trying to convince the FabFilter guys to implement minimum-phase oversampling as well as the current linear phase one. Sometimes this is preferred.

I have a weird habit of "hearing" the linear phase oversampling as a kind of added 'punch' or 'pressure' on audio material. It's like stuff sounds a tiny bit more 'hard' when oversampling is engaged on plugins. My guess is that it is the pre-ring that is somehow messing with the transients but this is just a guess. I've done numerous ABX tests and the results are still inconclusive but I just can't seem to ignore my gut feeling.

You should upgrade to the latest Elephant version though. Aleksey has implemented both linear and minimum phase oversampling modes in almost all his latest plugins and it really does make a difference. Mainly it's nice to have that choice.

All you Pro-L users.. please help me spam FabFilter about this so we get both options. I'd also like to have the option of much higher oversampling amounts. I've tried the plugin at up to 16 times oversampling using Christian Buddes plugin wrapper and Pro-L is untouchable at these rates. It just sounds ridiculously transparent at these high rates. :)

Cheers!
Niklas
Logged

nisilen

  • Newbie
  • *
  • Offline Offline
  • Posts: 3
  • Real Full Name: Niklas Silen
Re: Clipping & oversampling modes, audible differences
« Reply #13 on: April 11, 2011, 07:32:52 pm »

I don't use Elephant, but on Pro-L there is something that happens when switching to 2x and 4x that makes me very hesitant to use those modes... I'd like to know what's going on under the hood as well.

Here is my guess on what happens:

You hear the linear phase oversampling changing some of the internal parameters, causing them to be more 'accurate'. All internal envelopes will be much faster in operation and all filtering of the gain reduction paths will change slightly. All this should result in a much more transparent signal, that is, you get closer to the original source. There is definitely a very noticeable difference between running oversampling or none at all.

However, if it is bothering you, my guess is that you have the same issue as I do.. namely the linear phase oversampling implementation. There is a trade-off in having a perfect phase response up to nyquist and the inevitable pre-ring. It's kind of like the difference when running the Weiss EQ1 in linear phase mode versus it's minimum-phase mode. There's a subtle change in how the transients come through the unit and I usually prefer the minimum-phase mode. In my opinion causing some phase smearing of transients is a more acceptable trade-off than having perfect phase and pre-ring. It's especially noticeable on spikey or transient heavy material.

I really hope Frederic and Floris (fabfilter coders) take me seriously. I've been trying for a while to get them to implement a minimum-phase oversampling mode into Pro-L. The problem is that "on paper" the minimum-phase solution looks inferior and in reality the differences are incredibly small. Even after very thorough ABX testing and critical listening it might be hard to pinpoint the differences but they are there!! I'm not crazy! .. trying to convince them about this is the hard part. :P

Cheers!
Niklas
Logged

KAyo

  • R/E/P Forums
  • Full Member
  • *
  • Offline Offline
  • Posts: 237
  • Real Full Name: KAyo
  • Business Videos 24/7
Re: Clipping & oversampling modes, audible differences
« Reply #14 on: May 17, 2011, 04:47:43 pm »

The TC softclip is very useable.. you'd be surprised.
Give it a whirl.

Ciao'
KAyo
Logged
www.kantabiz.com
Business Video Directory
Pages: [1]   Go Up