Do ME's purposely clip?

[fenix]

I read in that Bob Katz TC manual that 3 clipped samples of percussion are usually inaudible. Look at the picture attached. Could it be that after limiting of the mix and done, some Mastering Engineers purposely raise the gain of the mix, say, 2db which would clip the wave? They're squeezing out 2 extra decibels and mostly inaudible distortion.

If you notice in the picture, I have zoomed in on a section of a song (a HOT song with an RMS of -7.5db!) with wavelab. The numbers on the top are samples. In the highlighted section, about 4 samples are flattened. Are they clipped? Or limited? If I had to guess I would say clipped. It takes some really really hard limiting to flatten like this.

 

[sonicpaint]

Clipping and limiting.

Mastering Engineers do a lot worse than that! Your showing a file that seems to still have at least 7.5db average to peak ratio, and that's quite good for todays "loudness wars". I have seen files that go as far as 3db average to peak ratio!

To answer your question as to if the file is limited or clipped? I think it's limited though if there is more than 3 samples hitting 0db at any given time it is also clipped, though it may not show on the meter.

Limiting these days is the norm in the mastering stage if the band or artist hopes to achieve the loudness of todays CD's being released commercially.

hope this helps,
sonicpaint

 

[kylen]

Yep, they certainly have the equipment and clients that pay them to do it !

I was chuckling about this thing someone posted the other day:
https://homerecording.com/bbs/showthread.php?threadid=103432

Broadcasters have had clippers for quite some time, I'm not that glad to see it creeping into professional audio I don't really get the point.

Of course when I clip something in my garage I just call it character !

kylen

 

[fenix]

No no no.

I'm talking about would they take the processed track and raise the gain of it. Clipping and limiting are 2 different things.

 

[kylen]

I'm talking about would they take the processed track and raise the gain of it.

Hi fenix,
I don't know if you're talking to me but I'll continue on just the same !

I don't have the link right now but clipping in broadcast is a good thing - they kindof do things differently over there to stay in the correct FM bands (heavy fines if they wander). There's pre-emphasis, AGC, limiting, clipping, and de-emphasis in your receiver.

Anyway the point of 'good' digital clipping at the sample level or limiting would be so that you can raise the gain (loudness) making the whole sound denser and whatnot - some folks call it squished.

You know all that, I've heard some of your stuff, Waves does that. You can either use a reasonable amount or too much. I'm not sure if Waves does sample level clipping, the stuff I have doesn't. You can test that out by trying to limit a burst - if there's overshoot then you don't have a clipper or limiter at the sample level. Lookahead found in digital limiters usually helps with attacks and overshoots.

Anyway that's what I found when I did some research and testing of what I had.

Is that the kind of stuff you're talking about?

kylen

 

[nessbass]

I think what he's talking about is what Bob Katz refers to as SHRED, and that's quite exactly what fenix said, i.e. raising the level after all the other processing has been done. This results in clipping and should ideally, if at all practised, be inaudible (3-6 full scale samples in a row).

But let me tell ya, luodness war is even worse than that. Take a look at RHCPs Californication, LOTS of shred. Or the Audioslave record. I once looked at some of the songs and there are kick drums with 30-40 samples in a row, flat as a ruler! And I won't have to tell you that that's very audible...

I hope this is going to be the last step in the loudness war, 'cause by now, the customers really get annoyed.

David

 

[nessbass]

Okay, I was looking for an example from Audioslave and zoomed in on one of the kick drums and saw this.

What the hell is that? Saw waves? Now, if some of the more knowledgeable persons could fill me in here...

Thanks.

BTW, the flattened part of the second positive pulse in the upper channel is 25 samples long.

 

[regebro]

Looks indeed like a nice low frequency saw + a square or triangle + square or something. Could be a kick, could be a bass, it's definiely a low and loud buzzing.

What is that scale in? Samples? seconds? Seconds make no sense, because then that triangle/square whatever would be some 6000 microseconds long, which is a frequency of 1.4k, which is NOT low. OK, so say samples. Then it's something down at around 62 Hz shaking your shpeekas!

 

[kylen]

fenix,
I read the spot in the Bob Katz article you're referring to. I think there are spots in there that are a little tricky...

Just before your spot he mentions something about most folks not being able to tell the difference between 10ms and 10us transient. Then in your section mentions a 3-6 sample percussion clip (overs) is usually undetectable. All I can say is try it out and see if that's the case for you. The clipping he's talking about is the bad stuff, digital overs, levels greater than 0dBfs.

I set my stuff to 1 sample over detection when possible - some apps I don't know what their overs threshold is (1 or 3 samples) before a red light gets turned on. At any rate I can always check it with Inspector (Elemental Audio).

Another tricky piece in the same paragraph is when he goes on to say that you can raise the gain by 2dB without using limiters or compressors. He says this right after talking about clipping and overs - and the heading of the paragraph is "Overs Counters and Increased Level". That's kind of an unfortunate to have all that in the same paragraph because digital overs are always bad and you may end up getting a reject from a CD manufacturing house because of it - not to mention you can hear that stuff too !

But when he says you can raise the gain by 2 dB without limiters and compressors he can't be seriously telling anyone to 'push into a clipper' and take 2 digital overs because chances are you can't hear them if they're percussion. That paragraph needs serious rework.

I hope he's telling us - if we don't use limiters and compressors to increase gain then just watch the 'overs' meter as you raise the gain or when you adjust the loudness by altering the EQ.

There's only 3 ways I'll ever adjust loudness in my garage - increase the gain by using a slider, increase the perceived loudness by adjusting EQ, increase the density by using compressor or limiter. Never by using overs...

kylen

 

[fenix]

Re: Re: Do ME's purposely clip?

I think what he's talking about is what Bob Katz refers to as SHRED, and that's quite exactly what fenix said, i.e. raising the level after all the other processing has been done. This results in clipping and should ideally, if at all practised, be inaudible (3-6 full scale samples in a row).

Yeah, this is exactly what i'm talking about.

Judging from looking at some waveforms, it would apear that this is what is occuring. Even if I take my L2 and slam it at ridiculous levels it won't flat top like digital clipping does.

 

[kylen]

Over on Bob Katz forum 'Shred' is used as a kind of slang the ME's refer to as over limiting or squishing.

fenix - are you talking about the software L2 or the hardware unit ?

I've got Cool Edit Pro and Sonar and Sound Forge but don't ever remember seeing flat-tops down at the sample level but I only had software plugs at that point. Ozone2 was my fastest software limiter with an attack time of 100us. Cool Edit Pro has a 'hard limiter' but I could make that thing overshoot on a transient.

Now I've got a couple of hardware digital limiters I might run the experiment again - oh I said I wouldn't do that didn't I ? Ha Ha -No that was 'overs'... whatever sounds good, eh ?

kylen

 

[fenix]

fenix - are you talking about the software L2 or the hardware unit ?

soft

 

[nessbass]

@regebro:

Yes, that scale is in samples. And it's, as I said, supposed to be a kick drum hit with bass and guitars going on at the same time.
BTW, that 'sawy' waveform starts right where you see it and also ends within that picture shown, so it's really just

up-down-up-down,

that's it.

To me it looks as though the waveform has been heavily clipped and those clipped parts have been twisted around....I dunno...but how could one achieve this? Looks pretty broken to me.

David

 

[kylen]

nessbass & regebro,
I'm totally confused by that picture - which app are you in just out of curiousity ?

I'm used to seeing complex audio produced by acoustic based instruments (could be amplified too) and vocals that looks like the stuff in this article:

http://www.261.gr/broadcastaudio.html

The stuff in your picture looks like something being modulated by a sawtooth oscillator like someone else said. I've got a synth with waveshapers that might look something like that ! Ha Ha What is that thing that made the wave do that I wonder ?

kylen

 

[kylen]

OK fenix, here's some independent confirmation from Roger Nichols (famous ME) ...I was looking for this article earlier.

http://www.rogernichols.com/EQ/EQ_2000_02.html

He's saying exactly what you and Bob Katz said - ME's will push past the digital ceiling a bit allowing overs (the most absolute digital limiter possible I would say) and then have the pressing plant back the volume down about .1dB or so to prevent overs (once the wave has been sawed off). That's exactly what your wave looks like, sawed off and backed down a bit.

Just as a caution - in case anyone is reading this and wants to try that at home. Ha Ha ! I recently re-balanced a project for a buddy of mine (in Cincinnati no less!) who had pushed across digital 0dBfs due to some technique issues. It's really tricky if you do that and in certain cases unrecoverable, in other cases you can get a cool sound (just make sure you don't send anything to a DAC with overs though - you won't like it !). Just keep a safety copy of your tracks incase downstream a bit you don't like the sound and want to remix. Best thing is to control yourself and stay under the overs while recording and mixing - there's other ways to get that sound.

As fenix found out though the pro ME's will do it to adjust loudness - Ha Ha must be a good thing, eh ?

kylen

 

[bleyrad]

it can be a pretty wise move to get things louder.

it doesn't pump at all like a limiter, and it can sound louder on the whole while still being pretty clean. of course, you can take it REALLY freakin loud if you don't mind dirty, distorted crap.


somebody mentioned something about the dup plant taking it down .1db to prevent overs... what?
i thought there was no such thing as an 'over' in the digital domain. 0db is as high as it gets.
there should be no need to back down to .1db, unless maybe it's being played on a very old and very flakey cd player that can't handle 0db.

 

[Outlaws]

Just my two cents from a standpoint of the layman musician who just fucks around and uses what sounds good............

Could that Audioslave pic listed a few posts back be a result of really hard limiting.......then during the mixing process they used the track fader on it to give it a drastic sound effect of coming in, then falling out, then coming back in?

I have done stuff where I will work the fader to the beat for a VERY strange effect that is very musical. ie coming in on the beat and fading out inbetween beats.

What seems logical to me is that the kick and the bass guitar line would have already be limited and set to a good level where they are flat tops and then mixed...the bass guitar is worked with the fader and then when the kick happens you get the sudden jump in the waveform.

 

[kylen]

somebody mentioned something about the dup plant taking it down .1db to prevent overs... what?

Hi bleyrad,
I'm one of the guys who mentioned that, right behind you. It's from the Roger Nichols article I reference that is a few years old by now but they site the same reason you do. Some older CD players would distort playing some music at 0dB depending on some other factors mentioned.

I think the 'overs' thing of a 0dB digital ceiling is simply because of DACs. I guess in discussions it comes up more often concerning media sent off to the mastering house or audio CD's burned.

Inside some of the DAW apps I have Like Cool Edit Pro there is no such thing as a digital over at 0dB when in 32bit floating point. In other words I can make a mix as hot as I like as long as I remember to bring it back down to within the limits of the media I am distributing on. Of course I can't monitor while it's that hot cause as soon as it hits a DAC - ka boom.

Since I haven't yet used my outboard digital limiters I am only used to pushing into something like Voxengo ElephantHQ or Ozone2 for loudness and setting the output to -0.3dB to -0.1dB to protect against overs (because of overshoot in the software plugs). That works for me and my distributions sound loud and clean. I can check it using Inspector if I need to.

I really don't have any good experiences yet with digital 'overs' and that type of clipping (digital brickwall limiter actually!) but this thread fenix started has gotten me to think about them a little.

I might try it out with my outboard limiters, like I said my software limiters/clippers all overshoot a bit.

As far as loudness goes I always make the song as loud as it needs to be - no louder.

kylen

 

[bleyrad]

Just my two cents from a standpoint of the layman musician who just fucks around and uses what sounds good............

Could that Audioslave pic listed a few posts back be a result of really hard limiting.......then during the mixing process they used the track fader on it to give it a drastic sound effect of coming in, then falling out, then coming back in?

I have done stuff where I will work the fader to the beat for a VERY strange effect that is very musical. ie coming in on the beat and fading out inbetween beats.

What seems logical to me is that the kick and the bass guitar line would have already be limited and set to a good level where they are flat tops and then mixed...the bass guitar is worked with the fader and then when the kick happens you get the sudden jump in the waveform.

i don't think so... i'm pretty sure that clip only represents an extremely small fraction of a second.
and anyway, there is no evidence it's getting "louder" then "quieter" at any point. Maybe if you looked at it like it was a volume envelope... but it's not. it's a waveform.


the only explanations i can see for that waveform are:

a) extremely coincidental sum of all involved waveforms forming a triangle-wave-like effect.

or

b) synths were used somewhere in there

 

[bleyrad]

one thing you have to realize about the guys doing this digital-clipping thing for loudness, is that they generally will:

- playback on an uber-high-end DAC

- take that analog output and feed it into an uber-high-end ADC at slightly over 0db to get the clips


i.e. they don't do it in software. they let the ADC do it for them. most go through this process for format-conversion anyway (instead of running a sample-rate converter, for example) and to run outboard effects.


it's not going to have quite the same effect if you just drive the peak lights in Wavelab or something. high-end ADC's handle clipping much better than cheaper ones or software.