Beginners guide to A/D clipping

[audiovisceral]

From many of the posts on here, I understand that in many cases, loudness in commercial mastering is attained predominantly through A/D clipping. I gather that clipping via this method, with a high end A/D converter can generate positive harmonic distortion that sounds more musical and impactful than compression (which alters vibe and dynamics), limiting (which can suck away impact), or ITB clipping (which just straight distorts).

From what I've read, in pro studios, the mix goes out of the sound card (D/A), and is processed by analogue eqs, compressors, and limiters. After this, it is somehow 'driven' to clip as it is re-inputted to the computer (A/D).

It's how this 'driving' process takes place that I don't understand. How do you feed the signal back into a sound card A/D at a level high enough to make it clip by the A/D, but not by the unit doing the feeding? For example, if the sound is coming out of a limiter and you set the limiter output to greater than 0, the limiter would clip the signal, not the A/D, would it not?

I can't get my head around it.

Why do I ask? Besides to add to my musical education, I'd really like to try, just to get a taste for the whole thing, an A/D clip on my budget gear. My sound card is a PCI M-Audio Delta 1010. Let's say I eq'ed, compressed, and limited ITB. I imagine it would be thereotically possible to and take the signal coming out of the Delta's outputs (D/A), boost it somehow (?), and then feed that signal right back into the Delta's inputs (A/D) and use them to clip. Again, however, I don't understand what would need to be inserted between these two points to elevate the signal level and force the A/D to clip.

Any help would be appreciated.

Also, I'm going to go ahead and surmise that the A/D on my Delta 1010 is probably not going to sound all that awesome clipping-wise (though I would pretty easily gamble it should still sound better than a clipping plug-in like the free G-Clip). Based on that, what sort of low-end/budget/entry-level A/D converters suitable for clipping currently exist, model and money wise? Obviously, I'm not asking what Sterling uses here ...

Thanks.

ps. *Please note this is not intended to be another 'how do I make my home masters sound like the pros?'. I am just hoping for some straight fundamentals on how it works and could, at least theoretically or crudely, be employed by a DIY home/hobby musician, for, if nothing else, a bit of fun.*

[lucey]

A gain control on something near the end of the analog loop turned up => clipping AD. If it sounds bad it is bad. If it sounds better than a limiter it is better. Both together in small doses are often best.

[Cellotron]

Originally Posted by audiovisceral

From many of the posts on here, I understand that in many cases, loudness in commercial mastering is attained predominantly through A/D clipping.

This is not entirely correct information. Many (if not most) pro ME's do not clip their ADC the majority of the time - and some never do. Some ME's when clipping also prefer to clip at a digital gain stage instead of at the input of the ADC. Others prefer to use various digital limiting algorithms (either in hardware or plugin), some of which in fact do not "flat top" to the same extent that digital clipping does at all.

For my own work I use all these methods when very high average level has been requested by the client. What individual or combination of techniques works best is entirely determined by the nature of the material on the track and the goals. Some techniques lead to more distortion, some techniques lead to more lost transients - you need to balance out what suits the track and aesthetic goals the best.

I gather that clipping via this method, with a high end A/D converter can generate positive harmonic distortion that sounds more musical than simply clipping ITB.

I don't think it has anything to do with the practice generating "positive harmonic distortion" - instead I think that some high peak transients are going by so quick that when the tops are flattened out the resultant square wave is so quick that it doesn't offend the ear as noticeably. Some ADC's have analog front ends which simply have more headroom (resulting in less distortion when pushed) than others. Still - as soon as you're overloading them there will in fact be some form of distortion happening - whether minor or not.

It's how this 'driving' process takes place that I don't understand. How do you feed the signal back into a sound card A/D at a level high enough to make it clip by the A/D, but not by the unit doing the feeding?

First off is just sensible gain staging - having a healthy but completely undistorted signal level through the entire process chain. With today's equipment this luckily is fairly easy to achieve. Next a lot of ADC's used in mastering have input attentuators on them - for my own setup all I have to do is turn the attenuator on my Mytek Stereo96 up so that the level at the input is going over what would be digital zero at it's output - and voila! - you have clipping at the input. What you have to do is monitor at the output of the ADC (I have a live loop back from my DAW to a seperate DAC so I can do this) and listen to hear at what level things start to noticeably distort so you can set this point properly. Usually clipping at the ADC works best when you're just chopping off the very highest transient peaks.

For example, if the sound is coming out of a limiter and you set the limiter output to greater than 0, the limiter would clip the signal, not the A/D, would it not?

I'm not sure what you're asking here -in general digital limiters are used after clipping at the ADC and not before.

Most analog compressors/limiters (and even a few eq's) have output gain controls so you can increase gain after their processing allowing you to send more signal to equipment after them.

Generally I never have to do this to have enough level to optionally overload my ADC - I usually just have to open up the input attenuator so that it goes from just green lights glowing to the red ones firing off occasionally.

Also, I'm going to go ahead and surmise that the A/D on my Delta 1010 is probably not going to sound all that awesome clipping-wise

My guess is that it (just like many converters out there - including the PT HD192) will sound like @ss when it is clipped. There's only one to truly tell though - and that's to run some tests yourself.

Based on that, what sort of low-end/budget/entry-level A/D converters suitable for clipping currently exist, model and money wise? Obviously, I'm not asking what Sterling uses here ...

Lavry, Mytek, Benchmark, Apogee - all make stuff that can work for this that isn't priced as premiumly as other options. You should expect to fork out at least around $850 for 2 channels unless you find a deal on something used though.

Anyway - I think the sound quality of your music will ultimately much better in the long run if you do not clip it. But as these kinds of disclaimers seem meaningless in this day and age - I hope the above "helps"

Best regards,
Steve Berson

[MASSIVE Master]

It's how this 'driving' process takes place that I don't understand. How do you feed the signal back into a sound card A/D at a level high enough to make it clip by the A/D, but not by the unit doing the feeding?

Not that I'm a big fan of the whole "clipping" thing (it's not the only way to skin a cat, although I admit guilt to some extent) -- Even with conservatively calibrated converters (say, -18, which seems fairly standard for many units) I can't think of too many pieces of analog gear that can't go well beyond full-scale before it clips.

Whether the gear passes a quality signal at that level is another story - Some gear does - There's a reason why many "mastering grade" (for lack of a better term) pieces are so pricey... Some gear doesn't.

Just because something isn't clipping doesn't mean it isn't distorting rather badly - Ask almost any cheap mic preamp or compressor that claims +24 but sounds like crap at +12.

[Lupo]

Clipping creates a waveform that is being programmed to overshot. What's being chopped off in the clipping, the output digital->analog converter will try to recreate.
(see this Nika Aldrich paper for explanation of the "hidden over" aspect of digital audio if needed)

The crux is that the output signal is bandwidth limited. Sudden fast changes can't exist anywhere else than inside the computer. A square peak in a row of digital audio sample points does NOT translate to a square topped wave on the output of the DAC. Rather the DAC will reconstruct the waveform peak back to something much like the level it had prior to clipping. In fact, most any processing involving filtering or bandwidth change, mp3'ing and so on, will recreate those 'hidden' peaks.

Try a change of bandwidth, say a sample rate conversion(SRC) from 44.1kHz to 192kHz. Depending on the nature of the clipping, this can easily create peak levels in the waveform that goes several dB's higher than the "brick walled" waveform initially indicates! Attached is an image of a typical loud master, before and after SRC from 44.1 to 192 kHz. Volume was lowered 6dB prior to processing. The sample peak level now indicates 2.5dB higher peak levels.

That's the beauty of clipping(and poor limiting) and probably the reason why many people thinks it sounds better than proper limiting. The clipped waveform doesn't loose all the peak level information. Those peaks are still existing in the digital audio stream, in a modified way, waiting to hit the consumer end where there hopefully is a 1000+ dollars state of the art digital->analog converter. Or not. Research by Nilsen and Lund at TC Electronics, available in the tech library on the TC website, indicates that most consumer CD player DAC's does not handle this abuse. In most cases, what sounds good IS good, but in this case - what sounds good only sounds good on expensive equipment. The master engineer have no possible way to know what's going to happen at the consumer end.

The problem is compounded when there is other processing involved, like a sample rate conversion or a psychoacoustic coding. In the picture example below, the gain was set at -6dB prior to processing to allow headroom for the new and extended peaks that 'suddenly' appeared after the SRC. In most real life conversions, like a consumer coding and decoding an MP3, there will be no headroom since the digital master is already set at maximum digital level. The result is distortion and more clipping.

So what's the solution? You could oversample the wave files to ridiculous degrees to get an idea what the reconstructed wave form looks like. Or.. The neat way: use an oversampled peak meter. If the sample rate is set high enough in the peak meter, it will indicate what's really going on at the output end of the system.

Using a normal PPM along with an oversampled PPM shows this relation between sample values and the real peak values directly. What is gained in the clipping is a direct consequence of pushing the information above the digital ceiling.


Andreas Nordenstam

[xqtion]

Wow, I feel like I just took a crash course in 'Audio Clipping'.

What a great info. I've been getting loud mixes using this method and it's really helpful to understand the process now.

Thanks Mr. LUPO!!

[Lagerfeldt]

Very well explained Andreas

[Bob Yordan]

Originally Posted by Lupo

That's the beauty of clipping(and poor limiting) and probably the reason why many people thinks it sounds better than proper limiting. The clipped waveform doesn't loose all the peak level information. Those peaks are still existing in the digital audio stream, in a modified way, waiting to hit the consumer end where there hopefully is a 1000+ dollars state of the art digital->analog converter. Or not. Research by Nilsen and Lund at TC Electronics, available in the tech library on the TC website, indicates that most consumer CD player DAC's does not handle this abuse. In most cases, what sounds good IS good, but in this case - what sounds good only sounds good on expensive equipment. The master engineer have no possible way to know what's going to happen at the consumer end.

So what happens if I clipp the signal at a lower level and in a 'controlled way', before the final stage? Will the peak information create any/less troubles for non
expensive equipments then to?

[flatfinger]

Regardless of how the results sound in the production environment, it's a crap shoot once the general public starts putting the end product into there cd players! Is'nt that some of the explanation behind whats happening??? The D/A and low pass filters at an M.E. 's place are going to re-construct more elagantly then those cheapy cd players that joe blow uses.

[lucey]

Originally Posted by flatfinger

Regardless of how the results sound in the production environment, it's a crap shoot once the general public starts putting the end product into there cd players! Is'nt that some of the explanation behind whats happening??? The D/A and low pass filters at an M.E. 's place are going to re-construct more elagantly then those cheapy cd players that joe blow uses.

I don't have a good technical answer, but speaking from my experiences listening to clipped masters in the car .... they actually sound better there than in here. The other distortions seem to mask the artifacts.

[dcollins]

Originally Posted by Lupo

Clipping creates a waveform that is being programmed to overshot. What's being chopped off in the clipping, the output digital->analog converter will try to recreate.
Andreas Nordenstam

How did you restore the peaks in that graphic?

DC

[Lupo]

Hi!

Thanks for the nice comments!

Originally Posted by Bob Yordan

So what happens if I clipp the signal at a lower level and in a 'controlled way', before the final stage? Will the peak information create any/less troubles for non expensive equipments then to?

If you clip in the analogue domain, before the ADC, all is fine. The AD will bandwidth limit the signal appropriately and an ADC, that is not driven into clipping, is guaranteed to make a stream of legal sample values/no hidden overs. As long as the signal is not tampered in the digital domain, for example normalizing. If you clip/hardlimit digitaly and lower the master level appropriately after clipping, there will be no 'hidden' overs either. But you'll gain no loudness.. and an A/B comparison between clip/non-clipped will quickly show you what sounds best. Even if the clipping by chance has it's charm for that particular piece of music, it's an unpredictable way of adding distortion. Most clippers and limiters are not bandwidth-limited, creating aliasing all over the signal. A test of a few typical limters, showing the aliasing of hard limiting and the benefits of oversampling the limiter, can be found here.

Originally Posted by flatfinger

Regardless of how the results sound in the production environment, it's a crap shoot once the general public starts putting the end product into there cd players! Is'nt that some of the explanation behind whats happening??? The D/A and low pass filters at an M.E. 's place are going to re-construct more elagantly then those cheapy cd players that joe blow uses.

Bingo!

Originally Posted by lucey

I don't have a good technical answer, but speaking from my experiences listening to clipped masters in the car .... they actually sound better there than in here. The other distortions seem to mask the artifacts.

Sometimes the other distortion masks the clipping, sometimes it adds up. Very hard to predict..

Originally Posted by dcollins

How did you restore the peaks in that graphic?

By upsampling from 44.1 to 192 kHz. The bandwidth was extended 4.35 times, so it's not entirely accurate as a way of representing the final output. Using a very steep filter at 20kHz would also account for the overshot of the DA output filter. In digital oscilloscopes, it's common to sample at at least ten times the signal frequency to give an accurate visual viev of the input wave. Oversampling to 440kHz is not an option in my system, nor 384kHz. Those who do have a 384kHz capable wave editor could use this to get a view of the final output wave.




Andreas Nordenstam

[absrec]

Allright. I've been waiting for an opportunity to ask about this and this seems like a great place/time.

I watched a friend of mine bounce down a mix one time. The mix was printed to a stereo track in PT. Before bouncing, he took the fader that the mix was on and turned it all the way up to +6 or maybe even +12! I looked at him strange and said "do you do that all of the time?". He said "Yeah, 'cause otherwise you have to turn your stereo up too much." I just shook my head and said "won't you get digital clipping without some kind of mastering limiter like an L1?". He said that this is the way he had always done it.

Next, we went out to his car and listened to it. I was sitting there ready to say "I told you so". Sure enough, it came on and I didn't hear any clipping! I was listening hard, too! It definitely sounded like a mastering limiter was catching the transients from the kick and snare, but it really sounded fine. And it had balls, too! I'm still scratching my head after that one.

I explained my thoughts on this to him and showed him the way I would bounce the mix. We used the only limiter he had available which was Maxim. I have more experience with the Waves L1 than Maxim, but I'm sure it was fine for our purposes. For one, we couldn't get it as loud without it sounding shitty. And the original "clipped off" version just sounded bigger. I agreed with him on that.

Anyone have any experience with this phenomenon?

-Aaron

[Cellotron]

Originally Posted by absrec

Allright. I've been waiting for an opportunity to ask about this and this seems like a great place/time.

I watched a friend of mine bounce down a mix one time. The mix was printed to a stereo track in PT. Before bouncing, he took the fader that the mix was on and turned it all the way up to +6 or maybe even +12! I looked at him strange and said "do you do that all of the time?". He said "Yeah, 'cause otherwise you have to turn your stereo up too much." I just shook my head and said "won't you get digital clipping without some kind of mastering limiter like an L1?". He said that this is the way he had always done it.

Next, we went out to his car and listened to it. I was sitting there ready to say "I told you so". Sure enough, it came on and I didn't hear any clipping! I was listening hard, too! It definitely sounded like a mastering limiter was catching the transients from the kick and snare, but it really sounded fine. And it had balls, too! I'm still scratching my head after that one.

I explained my thoughts on this to him and showed him the way I would bounce the mix. We used the only limiter he had available which was Maxim. I have more experience with the Waves L1 than Maxim, but I'm sure it was fine for our purposes. For one, we couldn't get it as loud without it sounding shitty. And the original "clipped off" version just sounded bigger. I agreed with him on that.

Anyone have any experience with this phenomenon?

-Aaron

Your friend was just doing simple clipping at a digital gain stage. Sometimes this can indeed sound cleaner than clipping at the input of the ADC. It can also sometimes allow things like snares and kicks to retain their sharpness better than the same amount of gain reduction as using a limiting algorithm. As always what technique works best for attaining high average level with minimum of artifacts will vary with the track.

Best regards,
Steve Berson

[Ecktronic]

Very interesting stuff.

Thanks all.

Eck

[Bob Olhsson]

Digital clipping doesn't sound that bad until you process the audio some more such as making an MP3 of it or playing it on the radio.

The problem is that you can't count on influential decision makers listening only to CDs on boom boxes or in cars. Some listen only to MP3s while others listen on really high-end systems.

The goal in all of this is to advance the artist's career. The goal of mastering is to make a great first impression on the industry types and a great lasting impression on the fans.

[Lagerfeldt]

Not to mention how clipped audio can sound on some radio stations.. pretty awful.

[Cellotron]

Originally Posted by Lagerfeldt

Not to mention how clipped audio can sound on some radio stations.. pretty awful.

Agreed! It's been posted a million times before but is worth posting again for all the newbies who think based on various internet mastering forum threads that somehow heavily clipping is now the "magic bullet solution" -

Frank Foti (Omnia) & Robert Orban (Orban) on
"What Happens to My Recording When it's Played on the Radio?":

http://www.omniaaudio.com/tech/mastering.htm

Best regards,
Steve Berson

[flatfinger]

Talk about a crazy situation!!! Freedom of speech but no freedom of the volume control!!!

amend the constitution: life liberty and the pursuit of happiness and the power to set the volume to your own taste!!!

[Cellotron]

Originally Posted by Leakycap

wow this is a great tip. I just tried it now. I printed my mix and pushed the fader to the top, I did it again, and then another time. my mix is super duper loud. Then I just lower the fader a little bit and now no clipping lights and it's still super loud. It got a little fuzzy though, how do I get rid of that?

The answer is simply backing off how much you are clipping it. Clipping by its nature induces some amount of distortion - the more you clip the more distortion you will get.

also the snare is very soft, can I just add a snare on top of that again?

You can do whatever you want - but my own practice would be to
instead go back to the original mix and bring up the snare.

but it's really loud otherwise. see it's tips like these that I like. I don't need to spend buckets of money on gear. with my laptop and some m-audio pro speakers, I can get my mixes as loud as anyone doing this for money. maybe i should charge for this . anyone here from the midwest area? how much do you guys charge for mastering?

Best regards,
Steve Berson

[Cellotron]

Originally Posted by flatfinger

Talk about a crazy situation!!! Freedom of speech but no freedom of the volume control!!!

amend the constitution: life liberty and the pursuit of happiness and the power to set the volume to your own taste!!!

You have that power no matter what the mastering engineer does.

You don't however have the power to add dynamics, transients or clarity back to a recording that has been smashed.

Main thing with smashed mixes: after the first few seconds where it sounds "loud" the end listener will adjust their volume knob. After this a smashed track can often be perceived as softer than a dynamic one.

Best regards,
Steve Berson

[flatfinger]

Originally Posted by Cellotron

You have that power no matter what the mastering engineer does.

You don't however have the power to add dynamics, transients or clarity back to a recording that has been smashed.

Main thing with smashed mixes: after the first few seconds where it sounds "loud" the end listener will adjust their volume knob. After this a smashed track can often be perceived as softer than a dynamic one.

Best regards,
Steve Berson

Ah yes, "wimpy loud"

[bonne]

Originally Posted by Cellotron

I don't think it has anything to do with the practice generating "positive harmonic distortion" - instead I think that some high peak transients are going by so quick that when the tops are flattened out the resultant square wave is so quick that it doesn't offend the ear as noticeably. Some ADC's have analog front ends which simply have more headroom (resulting in less distortion when pushed) than others. Still - as soon as you're overloading them there will in fact be some form of distortion happening - whether minor or not.

First off is just sensible gain staging - having a healthy but completely undistorted signal level through the entire process chain. With today's equipment this luckily is fairly easy to achieve. Next a lot of ADC's used in mastering have input attentuators on them - for my own setup all I have to do is turn the attenuator on my Mytek Stereo96 up so that the level at the input is going over what would be digital zero at it's output - and voila! - you have clipping at the input. What you have to do is monitor at the output of the ADC (I have a live loop back from my DAW to a seperate DAC so I can do this) and listen to hear at what level things start to noticeably distort so you can set this point properly. Usually clipping at the ADC works best when you're just chopping off the very highest transient peaks.

Best regards,
Steve Berson

Hi Steve,

Good stuff! I'm demo'ing the UA 2192 this week and one of the things I'm trying is different degrees of harmonic distortion -> clipping from the "colorful" UA front end for mastering projects.

A bit fiddly since it only has the tiny screwdriver-type attenuators round the back. The Mytek Stereo96 seems to be easier to work with in that respect. I also like the small footprint.

Up till now I've worked entirely ITB, but have felt the need for a round trip to analog for reconstruction of the signal. Also to avoid leaning too strongly on digital compressor/limiter for my level.

I saw in another thread that you contemplated exchanging your Mytek96 for the Lavry Blue for ADC back from analog. Do you find the Mytek lacking in any way in this application? How does the Mytek 192 compare? Is the front end identical to the 96? I understand the Mytek 8x96 has a different front end that doesn't work as well for this task.

Regards

Jørn Bonne

[Unclenny]

What a great thread........especially for a home studio guy like me who has always been content to slap on the old L1 and call it a 'master'.

I found that I was careful enough with my gain staging throughout this project that I could lose the limiter, crank my master fader to +4.5 and print out a fat, dynamic wave with just a touch of clipping that sounds MUCH better.

You guys are good.

[Fman]

Originally Posted by Unclenny

What a great thread........especially for a home studio guy like me who has always been content to slap on the old L1 and call it a 'master'.

I found that I was careful enough with my gain staging throughout this project that I could lose the limiter, crank my master fader to +4.5 and print out a fat, dynamic wave with just a touch of clipping that sounds MUCH better.

You guys are good.

+1. Ahh, to lurk.

[Cellotron]

Originally Posted by bonne

Hi Steve,

Good stuff! I'm demo'ing the UA 2192 this week and one of the things I'm trying is different degrees of harmonic distortion -> clipping from the "colorful" UA front end for mastering projects.

A bit fiddly since it only has the tiny screwdriver-type attenuators round the back.

In this case, instead of using the attenuators on the ADC you could alternately just turn up an analog gain stage at the end (i.e.something like an active line stage or if you are compressing the signal with an analog box just turn up the output gain on the comp).

The Mytek Stereo96 seems to be easier to work with in that respect. I also like the small footprint.

Me too - although half rack units instead of 1/3 rack space like the Mytek stuff doesn't bother me - plus the UA units has DA on it also, which kind of explains the larger full rack space size.

I saw in another thread that you contemplated exchanging your Mytek96 for the Lavry Blue for ADC back from analog. Do you find the Mytek lacking in any way in this application?

The Mytek just does straight hard clipping so you have to be careful not to overload it too much because once you go past just clipping the highest transient peaks you really risk distorting the material. The Lavry Blue ADC has a couple of optional "saturation" features - one analog, one digital - that may be useful when clipping inputs - and it also has a reputation of having slightly more headrom than most other converters. I haven't been able to test this out for myself yet though.

My main issue is that since my 4496 rack does not have an input sync module in it I would need to send it back to Lavry to have the ADC installed - so I have to figure out when I can afford the downtime due to not having my two best DAC's on hand.

How does the Mytek 192 compare?

Ergonomically they are almost the same.
Apparently they upgraded some analog components and the capacitors, they added the ability to record at the 4xFs rates, and they eliminated the 16bit recording and dither options. I have not heard the Stereo192ADC yet though so I don't know how it sounds in comparison to the Stereo96. It's possible I might get one in to demo later this year though.

I understand the Mytek 8x96 has a different front end that doesn't work as well for this task.

From my understandig the Stereo96 units had some minor improvements over the 896 in things like the AES receiver - but that essentially the converter circutry is exatly the same in both. The 896 does not have an adjustable input attenuator on its front panel though.

Best regards,
Steve Berson

[SixAndChange]

This is pretty sweet.

[SUPADUPA]

Crane Song Hedd 192 .... class A front end ...

[bonne]

Originally Posted by Cellotron

In this case, instead of using the attenuators on the ADC you could alternately just turn up an analog gain stage at the end (i.e.something like an active line stage or if you are compressing the signal with an analog box just turn up the output gain on the comp).

...

The Mytek just does straight hard clipping so you have to be careful not to overload it too much because once you go past just clipping the highest transient peaks you really risk distorting the material. The Lavry Blue ADC has a couple of optional "saturation" features - one analog, one digital - that may be useful when clipping inputs - and it also has a reputation of having slightly more headrom than most other converters. I haven't been able to test this out for myself yet though.

Thanks Steve!

For the moment I'm working ITB with high quality plugs, no outboard gear to drive the input stage of the UA. I'm coming straight in from my DAC. Therefore I'll have to fiddle with the input trimmers to experiment with the UA front end. Will try it out this weekend.

If I understand you correctly the Mytek96 does not allow for the gradual increase of good sounding gain like the UA front end does, but will rather quickly go into unwanted distortion. Right?

Cheers

Jørn