Mastering Without Clipping Converters

[lucey]

Originally Posted by jordanstoner

Bang - that's all I meant in my initial post. The definition of clipping is analogous to distortion, so I was confused when you mentioned "clipping without distortion." Perhaps a better interpretation is "clipping without audible degradation or artifacts"?

Well, he meant that to my reading. Almost everything causes some distortion or another.

Ah the internet ... we must be specific else we are wronged!

[sonicdefault]

We also have to keep in mind the potential for playback systems to reconstruct certain overages. I've learned some things about this subject awhile back when listening to Paul Frindle, but I'm no expert. I assumed some people in this thread may have had this quietly in mind. Anyone have any comments about this?


-SD

[thermos]

To those so totally appalled by the thought of clipping great converters for level, who of you have tried it? I think a lot of the disgust comes from the theory of it, without knowing what it really sounds like.


I clip my Aurora converters too for level when I master. I've tried every limiter available for mac, and absolutely hands down clipping is the most transparent way of getting level I've found. I don't go for super hot, around -13 rms or so, but even with modest reduction all limiters sound god awful to me while clipping sounds more than acceptable.

[fenderbender.]

Originally Posted by jordanstoner

Bang - that's all I meant in my initial post. The definition of clipping is analogous to distortion, so I was confused when you mentioned "clipping without distortion." Perhaps a better interpretation is "clipping without audible degradation or artifacts"?

No. Well, not any more than analog is analogous to distortion!

Clipping a Weiss and clipping a Lavry are two very different things.

And everybig mastering house is clipping their A/D. Like, every, as in... have you ever read that book called "Everyone Poops" ?

[MarkRB]

Originally Posted by fenderbender.

No. Well, not any more than analog is analogous to distortion!

Clipping a Weiss and clipping a Lavry are two very different things.

And everybig mastering house is clipping their A/D. Like, every, as in... have you ever read that book called "Everyone Poops" ?

No I haven't, is it good? Where the hell do you have to live to need to read a book on how to take a shit?

Clipping means overloading which equates to fatiguing annoying masters. Sure it makes stuff loud but does it really make the song nicer to listen too?
If i send a track to mastering I generally would like it to come back sounding better.

[lucey]

Originally Posted by MarkRB

Clipping means overloading which equates to fatiguing annoying masters.

that's just not the case if it's done well ... if you'd like, send me a mix and I'll clip it 'nicely' for you ... maybe even add some L2. limiting can be better or worse, it's another interesting part of the puzzle.

[Masterer]

Originally Posted by MarkRB

No I haven't, is it good? Where the hell do you have to live to need to read a book on how to take a shit?

Clipping means overloading which equates to fatiguing annoying masters. Sure it makes stuff loud but does it really make the song nicer to listen too?
If i send a track to mastering I generally would like it to come back sounding better.

Not necessarily. You've probably heard many records that were clipped that you loved. You can only tell it's happening if it's gone too far.

[soundroid]

how many classic albums have had the needles on the multi track pinged at 'max'.
heaps of engineers I know did this daily on the 'overhead' channels as well as others.
that was distortion wasn't it? the guys that did it simply knew what was musical and what was blatant overloading.

[Bang]

Originally Posted by sonicdefault

We also have to keep in mind the potential for playback systems to reconstruct certain overages. I've learned some things about this subject awhile back when listening to Paul Frindle, but I'm no expert. I assumed some people in this thread may have had this quietly in mind. Anyone have any comments about this?


-SD

yeah, stick an L2 on the channel you are recording to with a celing of -.03 with NO THRESHOLD. No overs...

[thermos]

Originally Posted by MarkRB

No I haven't, is it good? Where the hell do you have to live to need to read a book on how to take a shit?

Clipping means overloading which equates to fatiguing annoying masters. Sure it makes stuff loud but does it really make the song nicer to listen too?
If i send a track to mastering I generally would like it to come back sounding better.

To me fatiguing annoying masters is an L2 doing 1 db of reduction. My Auroras clipping 2-3 db sounds beautiful. Gotta try it before you knock it.

[Bang]

Originally Posted by thermos

To me fatiguing annoying masters is an L2 doing 1 db of reduction. My Auroras clipping 2-3 db sounds beautiful. Gotta try it before you knock it.

BINGO.

[macc]

I've been thinking about this a lot, and I can't fathom what the difference is between good-sounding clipping AD's and not. Hopefully someone can help fill in the gap in my understanding.

Now, I know that top-notch AD's have higher headroom analogue stages and so on (at least that's what you read about here), but nevertheless, that headroom is still over 0dBFS. This is what I don't get - how do such AD's 'translate' that over-zero information any better than anything during the conversion stage?

That is, in my limited (no pun intended!) knowledge, no matter how high the analogue stage's headroom, when it undergoes sampling/conversion (assuming it is 24-bit) it is still over 0 - ie flat-top. How can one flat top be different to another?

I'm interested in how this works, in what the difference is, not looking to exploit it. Clipping different ways with my rubbish little set up gives obviously different results - all pretty ugly - but assuming you aren't clipping an analogue stage, why should some digital clipping sound different to any other? What am I missing?


Cheers!

[arf]

Originally Posted by macc

but assuming you aren't clipping an analogue stage, why should some digital clipping sound different to any other? What am I missing?


Cheers!

IMO, you're not missing anything.

[macc]

Is that a trick answer? I think I thought it was, then I thought about it and now my head just hurts

[Lupo]

Originally Posted by macc

Now, I know that top-notch AD's have higher headroom analogue stages and so on (at least that's what you read about here), but nevertheless, that headroom is still over 0dBFS. This is what I don't get - how do such AD's 'translate' that over-zero information any better than anything during the conversion stage?

I don't really know, but I think it goes like this:

Clipping creates fast changes. Fast changes are high frequency. If this is done purely digital, those high frequencies have no where to go but to alias into the audio range. That's why clipping at higher sample rates often sounds better - less alias.

If the sampler clips prior to the antialias filter, the extraneous high frequency information will be removed and the clipping will sound cleaner.


Andreas Nordenstam

[grandmasters]

Originally Posted by Lupo

I don't really know, but I think it goes like this:

Clipping creates fast changes. Fast changes are high frequency. If this is done purely digital, those high frequencies have no where to go but to alias into the audio range. That's why clipping at higher sample rates often sounds better - less alias.

If the sampler clips prior to the antialias filter, the extraneous high frequency information will be removed and the clipping will sound cleaner.


Andreas Nordenstam

hmmm i think too much thought is being put into this. what is clipping converters? are we talking what you hear or what you see? if it's what you see then every converter interprets 'over' differently. My dCS will not say 'over' at what the Lavry will, yet I don't hear it on either. There is either distortion or there is not... and even if there is distortion... it can still be ok. Its always depends on the program material.

[thermos]

Originally Posted by grandmasters

hmmm i think too much thought is being put into this. what is clipping converters? are we talking what you hear or what you see? if it's what you see then every converter interprets 'over' differently. My dCS will not say 'over' at what the Lavry will, yet I don't hear it on either. There is either distortion or there is not... and even if there is distortion... it can still be ok. Its always depends on the program material.

Agreed. You guys are over thinking it, just try it and see if it works. And yes, clipping converters certainly doesn't work on all types of music/mixes.

And to macc, all converters do clip differently, probably because of the quality of the chip and the analog stages. Thats part of the appeal of the Lavry Gold/Pacific Microsonics Model 2 A/D is that you can get a lot more level out of them without hearing distortion.

[Lupo]

Originally Posted by grandmasters

hmmm i think too much thought is being put into this.

You mean that I have to.. listen??? Yikes!

[Hybrid]

Originally Posted by Lupo

I don't really know, but I think it goes like this:
Clipping creates fast changes. Fast changes are high frequency. If this is done purely digital, those high frequencies have no where to go but to alias into the audio range. That's why clipping at higher sample rates often sounds better - less alias.
If the sampler clips prior to the antialias filter, the extraneous high frequency information will be removed and the clipping will sound cleaner.
Andreas Nordenstam

You're right that the square waves created from clipping changes the frequency spectrum and that this causes severe aliasing back into the audible spectrum. The problem with clipping at higher sampling rates in studios and mastering is that as soon as it is made into a 16/44.1 file all the problems will surface.

The theory of digital sampling dictates that the frequency range going out of a D/A has to be the same as that coming in through the A/D - clipping changes this. It is because of the antialias/low pass filter that aliasing distortion and +dBFS signals are created. When this filter is moved up to 40kHz the aliasing will no longer be as audible but the +dBFS issue will still be there.

Unfortunately for most of the listeners all the studios and mastering studios have converters that can handle some amount of +dBFS signals - this means that even if you guys can't hear the distortion the people buying the CD will - and it's not subtle either.

It is without exeption a huge dissapointment to buy any new CD - this is why I have spent most of my money on buying CD's of the music I listened to in the 70's and 80's. I bought the Madonna (confessions of a dance floor), the two latest U2 CD's, No Doubt, Justin Currie, and they are unlistenable on any high-resolution loudspeaker system. They are only good for background music. Old CD's like Yello (Baby), Police (Syncronicity), Fleetwood Mac (Rumours) - you can really play them loud without obnoxious distortion threatening to tear your ears off - and the soundstage: there is DEPTH and the instruments has life.

I notised someone mentioned "inexperienced engineers" earlier in this thread but in my experience it goes like this: The higher profile artist the worse the sound. What's the explanation?

I also noticed someone talking about "radio ready" - that myth really should have died a long time ago. Bob Katz, Orban and TC Electronics all have written books and documents on this.

Because of the issues with digital sampling digital clipping should not be used at all - especially not to the extent you can find it on CD's today where the clip indicator is on all the time.

It is obvious that the record companies will NEVER be successfull in selling DVD-A and/or SACD if they don't get their priorities right - audio quality needs to come back on that list.

My opinion is that some of the classics like dark side of the moon, ziggy stardust, rumours, thriller, bad, HiStory, Baby are classics not only because of the music but for a large part also because of audio quality. People still buy these CD's - who will buy todays horrible sounding CD's in 30 years?

H

[Hybrid]

Originally Posted by thermos

And to macc, all converters do clip differently, probably because of the quality of the chip and the analog stages. Thats part of the appeal of the Lavry Gold/Pacific Microsonics Model 2 A/D is that you can get a lot more level out of them without hearing distortion.

So what about the people that buy the CD's you master? Do they all have a Lavry Gold's? What if they get 10% THD from their player?
This is exactly the reason why clipping should be avoided at all cost.

H

[thermos]

Originally Posted by Hybrid

So what about the people that buy the CD's you master? Do they all have a Lavry Gold's? What if they get 10% THD from their player?
This is exactly the reason why clipping should be avoided at all cost.

H

See fenderbender's post.

[macc]

Originally Posted by thermos

Agreed. You guys are over thinking it, just try it and see if it works. And yes, clipping converters certainly doesn't work on all types of music/mixes.

It's not whether it works that interests me especially - it's HOW it works, and from there why it works on some AD's and not others. It's the physics of the situation that interests me (as a physics graduate, albeit nearly 10 years ago). Simply that and no more. What is it that dictates how over-zero signals get translated differently.

I'm over thinking it for the sake of understanding, rather than for the sake of twisting it to get another 0.25dB

And to macc, all converters do clip differently, probably because of the quality of the chip and the analog stages. Thats part of the appeal of the Lavry Gold/Pacific Microsonics Model 2 A/D is that you can get a lot more level out of them without hearing distortion.

I know that much - it's why they clip differently that I'm intrigued to know. Somehow information arriving at the digital domain above zero gets interpreted differently in one case to another. 'Probably because of the chip' is a bit vague (sorry). Ah well, not to worry.

[Unclenny]

From the low end of town.......

I sent my last batch of tunes out and back through my lowly ADI-2's. In the process I pushed them up until the clips lights flickered. I'm not talkin' flat tops here....OK, maybe a few.

But to my humble ears it sure beat crankin' the old L1. I was able to push the faders up on my Finals to competetive levels (is Soundclick competetive?) without cringing (much).

[Lupo]

Originally Posted by Hybrid

Unfortunately for most of the listeners all the studios and mastering studios have converters that can handle some amount of +dBFS signals - this means that even if you guys can't hear the distortion the people buying the CD will - and it's not subtle either.

Totally agree. Have done that rant too many times here to bother anymore. Seems no one is interested.. :(

Beginners guide to A/D clipping

EDIT: that old post needs an update to include the effects of filtering..

[thermos]

Originally Posted by macc

I know that much - it's why they clip differently that I'm intrigued to know. Somehow information arriving at the digital domain above zero gets interpreted differently in one case to another. 'Probably because of the chip' is a bit vague (sorry). Ah well, not to worry.

Yeah, I don't know the science of it. It just seems to work for me in my humble little place (and most MEs in their not so humble places). Some day I'll expend the necessary energy to understand it.

[Bang]

Originally Posted by Hybrid

So what about the people that buy the CD's you master? Do they all have a Lavry Gold's? What if they get 10% THD from their player?
This is exactly the reason why clipping should be avoided at all cost.

H

This makes no sense at all. Once you have clipped via a high quality A/D and put a ceiling on it, the % of THD doesn't matter, its just playing back a digital file. A shitty playback system will make something squashed with an L2 sound just as bad as something overloaded to an A/D.

[MAzevedo]

Originally Posted by Bang

This makes no sense at all. Once you have clipped via a high quality A/D and put a ceiling on it, the % of THD doesn't matter, its just playing back a digital file. A shitty playback system will make something squashed with an L2 sound just as bad as something overloaded to an A/D.

If you have a digital signal with consecutive full scale samples, you will get intersample peaks greater than 0dBfs at the output of the alias filtering in the D/A converter. A cheap D/A may not have the analog headroom to handle the extra level, and will distort on playback.

[Bang]

thats why you put a CEILING on it. -.03 works.

[MAzevedo]

For heavily clipped material, you will get intersample peaks much greater than .03 dB. The "over" lights won't go off (I personally have my limiter ceiling at -0.3dB), but you can still clip the analog stage in a cheap D/A.

A good way to see this: Take a mix and clip it hard, then drop the level -0.1dB. You'll see no overs on you meters. Now, sample-rate convert it, up or down. The resulting files will have overs in it from the aliasing filter in the SRC. The same changes happen in D/A filtering as in SRC filtering, a clipped signal can and usually will have intersample peaks meaningfully higher than full scale.