Opinions; W1 Limiter and TLS Pocket Limiter

[ddummer]

Indeed.. Digital clipping is when you pass the top number
of the DA converter and it will go from a full plus to
a full minus value.(or the other way around..)

I don't know if any device these days would allow this...
You normally limit so you get that flat line att the top/bottom
instead of a wrap around....

//Daniel

[djsubject]

OT Question?

in the digital relme

is the number abouve the max number the minimum number?


carry on

Subz

[ddummer]

16 bit audio max number if you use unsigned short is 65535 then
the next number (one plus) will be 0.

signed short goes to +32767 and the next (one plus) will be -32768 so imagine the cone of a speaker doing that kind of
jump and you'll know what digital clipping would sound like..

//Daniel

[martian]

that is not digital clipping. digital clipping is when the DA convertor tries to output a number larger than is allowed and makes a horrid CRACK on the speakers.

everything else is limiting.

I'm pretty sure that's all LM was saying.

so you are saying digital clipping cant be performed with code?!

how after reading that wiki descption can you say that? other than building a gainer i dont imagine there is much thats simpler to code.

you also seem to miss the connection between hard limiting or saturation with no knee and clipping.

furthermore, LM described converting to 16bit as cutting off any signal above 0db, there we are on the same page.

One last try: a digtial clip ONLY occurs when digital information is truncated. 24- and 32-bits have so much headroom that this doesn't happen when you're working at those resolutions. Good luck digitally clipping a track by turning up the gain. It won't happen. And I know you agree, which is why you're now talking about limiters and other ways of "clipping" the waveforms in the digital realm (which aren't "digital clipping" despite being "clipping" and being done in the digital realm). However, when you convert to 16-bit for CD, any of those 'bits' that are above 0dBu are now simply cut off, resulting in a nasty and unusable (except as an effect) sound. It is lost information, which results in a crackling distorted sound. That's gotta be the last time I try to explain it, or I'm going to go nuts.

Whatever it is that you're referring to (the wave being shaped by a limiter or whatever), it's NOT "digital clipping".

but if i provide drum samples driven 4db into the master and exported at 16bit, he will tell me its not digital clipping. it seems the only time it can be described as 'digital' clipping is when it sounds bad.

[c_huelsbeck]

No matter what the exact definition, I still say it is a bad idea to use this... it may be a quick and dirty way to change the sound of a drumloop to the hobby producer perception of "phat sound", but it opens a can of worms for various reasons when, at the same time, the same phatness could be achieved by good engineering and not bending physics out of shape.

Good reads that I found:

http://www.tcelectronic.com/media/Level ... AES109.pdf

http://ff123.net/norm.html

[martian]

yes, the whole tc tech archive is very informative. similar topics in BKs 'mastering audio' too.

although the mp3 thing is of big concern currently, it may only be tempory with the rate of storage growth. generally though i find the smaller the peak to rms ratio of the master the more peaks and the higher they will occour from mp3 compression, with them getting worse the lower the bitrate. some loose slightly unpredictable relationship like that is what ive found.

incidenty one of the causes of peaks (bandwidth limitation/filtering) i tried to point out to greg here: http://www.kvraudio.com/forum/viewtopic ... ilter+peak

[defjamm]

No matter what the exact definition, I still say it is a bad idea to use this... it may be a quick and dirty way to change the sound of a drumloop to the hobby producer perception of "phat sound", but it opens a can of worms for various reasons when, at the same time, the same phatness could be achieved by good engineering and not bending physics out of shape.

Good reads that I found:

http://www.tcelectronic.com/media/Level ... AES109.pdf

http://ff123.net/norm.html

it has nothing to do with phatness or good engineering, if u want your hip-hop-track loud AND punchy you'll have to cut the tops of your kick and snare, be it in the mixing stage or in mastering, but there is no other way to do it. because then you can turn the rest up. you simply have to find the best way to do it, and the best way used in mastering today is to put the mix out into analog, raising the gain and simply clip your AD-converter, when it comes back into digital.
you'll need great converters for this.

you can look at the wave-forms of modern songs in an audio-editor, the top is cutted.

a cheaper way to do this(and the only if somebody asks for advice at kvr) is to use something that emulates this in the plug-in-world, like the timeworks-mastering-comp or the t-racks clipper or the clipping-mode in elephant or in ozone 3(i like the timeworks one the most for this) but NOT something like the w1, a waves l1-emulation. if you want to use ONLY a traditional limiter like the l1 to achieve your loudnedd AND punch you will fail, it's not possible.

load a modern loud song in your audio-editor of choice and look at the top of the wave, it'll be simply cut at the top where the snare and kick are playing.
only after the cutting they were lowered in gain, so that the wave doesn't clip.


@martian: the speed affects the soft-mode i believe. i simply leave it at 1 and use hard-mode.

[defjamm]

http://gearslutz.com/board/showthread.p ... 358&page=2

this thread is about sterling, one of the most-important mastering houses worldwide. chris athens is an engineer there and participates in this thread, really interesting stuff. you can read directly from a mastering-pro how they do it.
at some point they have to get rid of the peaks of the snare and kick without losing the punch and he describes how.

[Lunch Money]

Ah, still angry, then, Martian? Remember, anger is the child of pride!

I read the above referenced thread, and to be completely honest it's never clear to me which stages are analog and which are digital for any one person, since the path of one project is never really made clear. So many questions, with good answers, but on a variety of subjects some of which are not necessarily overlapping. Some people are concerned about getting "overs", while others aren't. Are they in 16-bit or 24-bit?, which is a hugely important factor for understanding ANY of it-- because their "overs" might not be clipping the waveform at the digital level. Some are talking about L2 hardware (didn't know such a think existed) while others are talking about L2 by Waves. Other than the person mentioned above, the expertise and understanding of ANY of the others isn't entirely clear. Indeed, even HIS expertise isn't really clear to me, though he seems no fool.

All I'm saying is that it's hard to tell who in that thread actually knows what they're talking about, which equipment is being used, at which stage the sound is being limited, and in what context the mastering engineer is fine with his "overs". Is "hitting the AD/DA" still an analog stage? I would think so-- the card doesn't just magically turn everything digital, or all converters would sound identical. The card has to decide how to make something digital, and some cards will do it differently than others. Seems to me that ddummer was describing this input stage as limiting, not clipping. You can't forget about the "analog" part of "analog to digital". These are all important factors, none of which change my basic premise which is:

Limiters will shape the wave so that the resulting waveforms are able to fit in the required word-length of 16-bit sound files. Digital clipping at the same bitrate is a nasty and unusable mess, which is only forgivable (and I've said this three times already... it's getting tiresome...) if the material is busy and/or percussive enough that the god-awful mess that is digital clipping is masked by the overall sound.

For crying out loud, it's self-evident. Companies wouldn't even MAKE limiters if all you needed to do was push your waveforms into overs, and people wouldn't discuss which ones accomplish their task in a more "transparent" way. This elusive transparency refers to developing sophisticated (I'm guessing... who really knows, maybe someone's chuckling about how easy they discovered their limiter's algo) algorithms to deal with re-shaping the waves so that they maintain their original character... to avoid, and not encourage, clipping.

As for the peaks thread, you didn't "try" to show me anything about resonant filter peaks in that thread, you DID show me. And then I didn't continue the point because-- hey, fair game-- what you said looked correct and was confirmed by Aleksy. Aleksy's mentioning that such peaks are pretty much negligible was confirmation that I wasn't completely out to lunch, even though it turns out that there are peaks of some sort. Was I SUPPOSED to respond again to that thread?? It seemed resolved.... I would have passed the rest of my life not even remembering it, so it's kind of curious that you do! I don't understand how my not bumping that thread is now somehow supposed to make me look bad...? Or are you trying to relate two completely unrelated points...? Not sure what that reference was all about.

Back to the digital clipping-- for crying out loud, I posted a freaking sound sample with clear audible evidence of what digital clipping does. What more do you want me to do? I provide a description, I agree that it might not always be noticeable depending on material, and I provide an example of what I'm talking about for the "horrific" sound of digital clipping. My logic adds up, my example demonstrates correctly, and yet I DID admit that it might not "always" result in an unusable track. Not sure what more you wanted from me, because I simply cannot and will not agree that a limiter is a digital clipper, or that digital clipping is a preferable solution for adding "punch" to a mix. A digital simulation of analog clipping, sure! But actual digital clipping? No thanks.

Greg

[c_huelsbeck]

you can look at the wave-forms of modern songs in an audio-editor, the top is cutted.

Right... that's why I can't stand most modern songs... it just sounds like crap!

[c_huelsbeck]

http://gearslutz.com/board/showthread.p ... 358&page=2

this thread is about sterling, one of the most-important mastering houses worldwide. chris athens is an engineer there and participates in this thread, really interesting stuff. you can read directly from a mastering-pro how they do it.
at some point they have to get rid of the peaks of the snare and kick without losing the punch and he describes how.

Good, another mastering house that I can strike from the list... just like that guy who maintains that some HDs sound better than others and even bit-identical transfers can sound different... never ever would I trust guys like that with my music.

[defjamm]

@lunch money: the hardware l2 is also from waves.

what i understand from him is, that he basically doesn't care about overs as long as it sounds good to him.
it wouldn't surprise me, if the best mastering engineers don't care much about such things like intersample-clips.

greg i repeat myself but you guys mean really the same thing, you just argue about the correct name for it. the audio-example you posted, it's not what he meant with digital-clipping, the example really sounds horrible and nobody would use that for modern music. it's just what you call the simulation of analog clipping, he calls it digital clipping.

[Lunch Money]

Yeah, I said pages ago that he's probably talking about something other than digital clipping.

If someone says, "I'm pretty good with my strum-banger," and I respond, "your what?" and the person says, "you know, my strum-banger. It has 6 strings... tuned EADGBE... you can play all kinds of music with it..."

I would say, "Oh, you mean a guitar."

(or an "axe"! )

Digital clipping is a specific enough phenomenon that it's very important to understand the difference between digital clipping and what he may or may not be describing (which is becoming increasingly unclear). At every turn, I've tried taking the opportunity to say, "Ah, but what you just described isn't digital clipping" or whatnot, but to no avail. Once, when someone (you, perhaps? I can't remember!) suggested we were talking about the same thing, he said something blatantly incorrect, and my response was, "OK, we might be thinking of the same thing, but that doesn't match up with what he just said," (paraphrased), so it's been tough to figure out anymore whether we even ARE talking about the same thing, or whether he's sharing (intentionally or otherwise) false information.

Greg

[martian]

Ah, still angry, then, Martian? Remember, anger is the child of pride!

that was a note to self as well as you, and my reason for quitting at that point. but you still threw more coal on the fire, i was willing to accept differences and move on.

Seems to me that ddummer was describing this input stage as limiting, not clipping. You can't forget about the "analog" part of "analog to digital". These are all important factors, none of which change my basic premise which is:

Limiters will shape the wave so that the resulting waveforms are able to fit in the required word-length of 16-bit sound files. Digital clipping at the same bitrate is a nasty and unusable mess, which is only forgivable (and I've said this three times already... it's getting tiresome...) if the material is busy and/or percussive enough that the god-awful mess that is digital clipping is masked by the overall sound.

For crying out loud, it's self-evident. Companies wouldn't even MAKE limiters if all you needed to do was push your waveforms into overs, and people wouldn't discuss which ones accomplish their task in a more "transparent" way. This elusive transparency refers to developing sophisticated (I'm guessing... who really knows, maybe someone's chuckling about how easy they discovered their limiter's algo) algorithms to deal with re-shaping the waves so that they maintain their original character... to avoid, and not encourage, clipping.

yes digital clipping is horrible on most stuff, but what i have said from the beginning is how different it is with heavy transient drum based material.

As for the peaks thread, you didn't "try" to show me anything about resonant filter peaks in that thread, you DID show me. And then I didn't continue the point because-- hey, fair game-- what you said looked correct and was confirmed by Aleksy. Aleksy's mentioning that such peaks are pretty much negligible was confirmation that I wasn't completely out to lunch, even though it turns out that there are peaks of some sort. Was I SUPPOSED to respond again to that thread?? It seemed resolved.... I would have passed the rest of my life not even remembering it, so it's kind of curious that you do! I don't understand how my not bumping that thread is now somehow supposed to make me look bad...? Or are you trying to relate two completely unrelated points...? Not sure what that reference was all about.

that was simply demonstrating your initial confidence in what you dont know and unwillingness to test/understand/accept things.

so you trust aleksey's opinion? pmed.

Back to the digital clipping-- for crying out loud, I posted a freaking sound sample with clear audible evidence of what digital clipping does. What more do you want me to do? I provide a description, I agree that it might not always be noticeable depending on material, and I provide an example of what I'm talking about for the "horrific" sound of digital clipping. My logic adds up, my example demonstrates correctly, and yet I DID admit that it might not "always" result in an unusable track. Not sure what more you wanted from me, because I simply cannot and will not agree that a limiter is a digital clipper, or that digital clipping is a preferable solution for adding "punch" to a mix. A digital simulation of analog clipping, sure! But actual digital clipping? No thanks.

Greg

clipping doesnt add punch. it takes it away like almost any transient reduction, but it will preserve it better than a limiter. its simple to understand, consider on one limiter (say elephant) why fast limiting preserves punch better than slow limiting, it also has to introduce more distortion too.

hypothetically there is a point where a limiter becomes a clipper, limiting is simply a form of saturation.

[martian]

Digital clipping is a specific enough phenomenon that it's very important to understand the difference between digital clipping and what he may or may not be describing (which is becoming increasingly unclear). At every turn, I've tried taking the opportunity to say, "Ah, but what you just described isn't digital clipping" or whatnot, but to no avail. Once, when someone (you, perhaps? I can't remember!) suggested we were talking about the same thing, he said something blatantly incorrect, and my response was, "OK, we might be thinking of the same thing, but that doesn't match up with what he just said," (paraphrased), so it's been tough to figure out anymore whether we even ARE talking about the same thing, or whether he's sharing (intentionally or otherwise) false information.

Greg

hows about you tell me how i can achieve digital clipping then?
everything i have describle as digital clipping is doing this:

In digital signal processing, clipping occurs when the signal is restricted by the range of a chosen representation. For example in a system using 16-bit signed integers, 32767 is the largest positive value that can be represented, and if during processing the amplitude of the signal is doubled, sample values of 32000 should become 64000, but instead they are truncated to the maximum, 32767.

ie. cutting the tops and bottoms off a waveform with no knee (aka infinitely hard) to the ceiling.

are you confusing 'wrapping' to digital clipping?

Clipping is preferable to the alternative in digital systems — wrapping occurs if the digital hardware is allowed to "overflow", ignoring the most significant bits of the magnitude, and sometimes even the sign of the sample value, resulting in terrible distortion of the signal.