Is 0db clipping?

[Jouni]

Hi.

There's some mixing/mastering attempts flying around of a couple of records I play in. And as I threw one mastering to Goldwave, it kept blinking the red lights. I let the poor mixercandidate have it.
Later I dropped the same mix into Wavelab and noticed that the red clipping-indicators did not blink. Apparently GW blinks on 0db and Wavelab does not.

Personally, I usually set my limiters to -0,3 db.. always done so just to be on the safe..

 

[leaningpine]

In digital yes, 0db is the highest you can go and you can't go one smidgen over that. I would recommend -3 as the highest to go.

 

[tigerflystudio]

Interesting. And to further this discussion....

In my DAW, if I wanted to (which I don't) I can actually push levels up to +5dB without any audible distortion or clipping whatsoever (distortion / clipping occurs at about +6dB or higher).

So, I guess what I'm asking is, what's worse for the audience (listener)?..

1) a low volume mix that peaks at (or just under) 0dB?
2) a mix than hovers around 0dB with occasional peaks of, say, +3dB?
3) a loud / compressed mix run through a Limiter that peaks at 0dB?

What are the pros & cons of these? Which is best? If my mixes aren't audibly distorting, is it wrong to go with option 2 above?

 

[SouthSIDE Glen]

Hi.

There's some mixing/mastering attempts flying around of a couple of records I play in. And as I threw one mastering to Goldwave, it kept blinking the red lights. I let the poor mixercandidate have it.
Later I dropped the same mix into Wavelab and noticed that the red clipping-indicators did not blink. Apparently GW blinks on 0db and Wavelab does not.

Personally, I usually set my limiters to -0,3 db.. always done so just to be on the safe..

0dBFS in and of itself is not digital clipping. Digital clipping is when you have a signal that tries to exceed 0dBFS, but cannot, so the top of the signal gets "clipped off" (hence the name).

As far as why the clip indicator goes off in one editor and not the other, I *just* explained this yesterday here.

G.

 

[tigerflystudio]

Hope I'm not hijacking this thread..

SouthSide, thanks for link.. useful info, as ever. Sorry to be a noob, but does that mean I should abandon how I am currently mixing (i.e. option 2 above) and instead start doing 1 or 2?

 

[RAMI]

does that mean I should abandon how I am currently mixing (i.e. option 2 above) and instead start doing 1 or 2?

If you're in 24 bit digital, there's no reason to be anywhere near odb before mastering. It has absolutely no benefit.

 

[tigerflystudio]

A lot of my tracks are 24-bit, recorded at around -5dB (peaks of -2dB).
Alas, before I decided to go 24-bit, I did a couple of tracks in 16-bit only. They are recorded at around -3dB, with peaks of -0.5dB.

(Probably just through sheer inexperience) I have always thought it best to record as close (upto) 0dB as possible, but never above it. Typically the peaks are infrequent and are a result of a couple of instruments combining to push the stereo mix over 0dB.

So, I was wondering if the solution would be to run mixes through a Limiter (to 0dB) to control / avoid any spikes going above? Or better to avoid a compressing / Limiting and to just drop the level of the track as a whole so the peaks are lower than 0dB?

 

[mshilarious]

WL won't show a clip on the master section unless you exceed 0dBFS. If you use the analysis tool, it will define a clip as four subsequent 0dBFS (+ or -) samples.

 

[SouthSIDE Glen]

So, I guess what I'm asking is, what's worse for the audience (listener)?..

1) a low volume mix that peaks at (or just under) 0dB?
2) a mix than hovers around 0dB with occasional peaks of, say, +3dB?
3) a loud / compressed mix run through a Limiter that peaks at 0dB?

What are the pros & cons of these? Which is best? If my mixes aren't audibly distorting, is it wrong to go with option 2 above?

What is "worst for the listener" is kind of a subjective question, but IMO, the answer is that #2 is worst, though #3 can be worse if it is over-done.

The thing about #2 is that the only reason you are even able to peak in positive numbers is because your DAW software is using floating point math, which allows that. But when it comes time to master your mix to either standard CD/CD-R format or to MP3, those formats use fixed math representations which cannot go above 0dBFS. So any peaks in your DAW that rise above 0 will wind up being clipped off at 0 . the only way to avoid that would be to lower your overall mix level, in which case we're back to #1 anyway, or to compress or limit it first, in which case were just doing #3.

I agree with Rami, that there's no reason why your mix needs to have peaks that high to begin with. Keep the mix clean and keep the peaks below 0dBFS, IMHO. If you absolutely *have* to pump up the volume, a little bit of #3 as part of mastering is OK. But so many people over-do that these days, that it can be tempting to over-do it yourself, just to try to keep up with the Jonses. Do yourself a favor, and screw what the Jonses are doing, and don't compare your volume to theirs. Instead listen to *your product* and don't push it so hard that you're sacrificing sound quality for loudness. And even more so, if you can't tell the difference, then leave the mastering to someone else.

G.

 

[Massive Master]

I don't mean to throw [SELF-CENSORED] on the windows here and I know that a couple people have already 'subliminally' throw it already with limited results -

But can we PLEASE specify what scale we're talking about - all the time? Or at least once in a sentence so we have a reference scale?

A mix that peaks at +3dBVU is fine.
A mix that peaks at +3dBFS is completely unacceptable.
A mix that peaks at +3dBSPL is nearly silent.

A mix that peaks at "0dB" is - Unknown.

"0dB" on its own, means nothing. Literally - "No Decibels" - Granted, that's the one thing that can be an assumption of sorts -- When people generally say "0dB" they're usually referring to unity gain (+/-0dB). You can also *ASSUME* that when someone says "-0.3dB" that they mean "-0.3dBFS" (but it's still an assumption - and you know what happens when people assume...).

Not to be picky - Just to be clear.

 

[mixsit]

..So, I was wondering if the solution would be to run mixes through a Limiter (to 0dB) to control / avoid any spikes going above? Or better to avoid a compressing / Limiting and to just drop the level of the track as a whole so the peaks are lower than 0dB?

And just to pick at this one, let's say we simply differentiate between the use of compressors/limiters (in recording) when needed, for the sound as opposed to when they are used in a strict sense 'as safeties'.
Very often the limiter can be said to be there on a mix (or track) 'to catch strays' but not for the sound of it' -but at that point it is to allow the sound to be raised to the point of it being needed.
Then there is limiting for a)catching 'strays but not audibly effecting the sound (other than enabling higher level), or b) more aggressive limiting that degrades as an acceptable trade off to higher level, and c) limiting for the effect of enhancing the sound.
Now you get to pick, which, when.

 

[flatfinger]

if you're cd is peak normalized to 0dBFS ( after a nice brick wall limiting session of course) you will have the warm and fuzzy of knowing that none of those precious ,precious digital bits have gone unused . By all means; Use every dam bit ! ( of course , I'll not be listening to the distorted results anytime soon!)

 

[suprstar]

But can we PLEASE specify what scale we're talking about

He means dbFS.

 

[SouthSIDE Glen]

I don't mean to throw [SELF-CENSORED] on the windows here and I know that a couple people have already 'subliminally' throw it already with limited results -

But can we PLEASE specify what scale we're talking about - all the time? Or at least once in a sentence so we have a reference scale?

A mix that peaks at +3dBVU is fine.
A mix that peaks at +3dBFS is completely unacceptable.
A mix that peaks at +3dBSPL is nearly silent.

A mix that peaks at "0dB" is - Unknown.

"0dB" on its own, means nothing. Literally - "No Decibels" - Granted, that's the one thing that can be an assumption of sorts -- When people generally say "0dB" they're usually referring to unity gain (+/-0dB). You can also *ASSUME* that when someone says "-0.3dB" that they mean "-0.3dBFS" (but it's still an assumption - and you know what happens when people assume...).

Not to be picky - Just to be clear.

You already know that I agree completly with you on this point, John.

I personally let it slide here because the Op was specifically talking about readings in Goldwave and Wavelab, which almost by default means he was talking about dBFS.

But you're right, that it's possible - it has happened many times in the past - that the natural evolution of a thread caused others to be talking about different scales (let's not forget dBu and dBV and others while we're at it.) Tigerfly's comments, for example, could just as easily be interpreted to be talking about dBVU as dBFS. I admit that I missed that one.

Unfortunately do many newbs to recording don't even know that there *are* different dB scales, let alone how they relate (or don't) to each other. This is, unfortunately a root cause of many rookie problems.

It's like people saying that it 25° outside. If you don't specify °K, °C or °F. you don't know whether that means that the sun ran out of fuel, that it's bikini and sunscrean weather, or below freezing.

G.

 

[timkroeger]

if you're cd is peak normalized to 0dBFS ( after a nice brick wall limiting session of course) you will have the warm and fuzzy of knowing that none of those precious ,precious digital bits have gone unused . By all means; Use every dam bit ! ( of course , I'll not be listening to the distorted results anytime soon!)

I know you didn't mean it but I have to disagree nevertheless I am no pro in the recording area but I read an article about intersample peaks in sound & recording magazine recently and did some googling. I found another simple article about the problem of intersample peaks. It's quite interesting and possibly no news to the pros but I always thought I'd be safe with 0dBFS, so now I'm wiser You might not notice a problem mastering at 0dBFS and most of the people with professional (or just "good") DA converters won't notice anything either but most low cost consumer gear clips on those intersample peaks. There are quite a lot of commercial CDs out there which sport intersample peaks and thus pose some challenge to low cost DA converters.

Joini, I'd say either just don't master for 0dBFS but for -1dBFS to avoid that but that costs you precious resolution OR just use a plugin like X-ISM from SSL that can show intersample peaks. That way you're pretty safe and you can set your levels and use limiters where you need to and you're not forced to use a limiter at the end of the chain just to be on the safe side. Btw. even if you use a limiter set to 0dBFS or -0.3dBFS just before the output, you're not safe from intersample peaks. You don't know if -0.3dBFS is sufficient to prevent them and neither do I.

Cheers
Tim

 

[mshilarious]

WL has an intersample peak meter somewhere . . . I think it's in bit meter (not at home at the moment). And any good peak limiter would be programmed to avoid intersample peaks.

 

[timkroeger]

I don't know when the conversion from floating point to fixed point math happens or if that matters at all. I think the "good" peak limiter you're referring to would then need to emulate a DAC and do some oversampling to analyze where those peaks are.

Do you know a peak limiter that does avoid intersample peaks for sure? I'd be inclined to think that's no trivial task and only needed under special circumstances (e.g. mastering to CD or mp3). From what I've read SACD is another story altogether. Judging from commercial CDs, a lot of mastering engineers seem to be using the wrong tools. In addition to that, mp3 (due to being a lossy codec), is even more prone to generate intersample peaks than CDs.

Using an inter-sample precision digital limiter with the threshold value set to -0.3 dBFS, that is, basically just to reduce overshoots, the distortion of the DACs disappeared almost completely. The overall loudness of the signal was not reduced, however, so for many applications this seems to be the better option.
Applying an inter-sample peak limiter before the perceptual codecs resulted in less clear results: The MPEG Layer II codec reduced the peak level both overall and the maximum peak value when the input signal had been treated with the limiter. The average size of the peaks was also reduced in the Layer III case, but certain of the limited input signals generated a few higher peaks than when the input signal had not been limited. This was surprising to the authors and shows the difficulty as estimating exactly what happens downstream in a signal chain and taking appropriate action.

I wouldn't want my peak limiter to care for those intersample peaks if I'd be mastering to tape because I know my DACs can handle the signal due to enough headroom. The same would go for SACD. In my opinion treating intersample peaks makes only sense in an "output format context", so it should happen on "transfer" instead of mastering.

I'm speaking all theoretical though, since I'm lacking a lot of real world practice & experience. I've just read a few other really interesting articles and just want to spread a little "awareness". It's an interesting subject and many people seem to be doing at least some things "wrong".

The following papers are quite techy but definitely worth a read. Basically just a few pages

• Digital Distortion in CD’s and DVD’s (Aldrich)
• Stop counting samples (Lund 2006)
• Overload in signal conversion (Nielsen & Lund 2003)
• 0dBFS+ Levels in Digital Mastering

Cheers
Tim

 

[Massive Master]

He means dbFS.

I knew that. Then a later post was referencing positive levels. Then a later post was talking about tracking levels.

Just being picky for posterity.

 

[miroslav]

Is 0db clipping?

No....it just means you're not trying hard enough!

 

[tigerflystudio]

Thanks for all the useful info here guys - really interesting reading. Sorry, I was referring to dBFS. I should been clearer - just thought is was a 'given' & didn't want to insult your intelligence. But fair point. My bad.

I use to get handed a lot of CD's from bands / artists who wanted to play at my club. I always listened to every one very carefully - both from a 'can these guys do-the-business' point of view, and also that of a budding home-recordist.

One thing that always struck me about the different recordings was the range of volumes - from mouse-quiet to bleeding-ears loud. I was forever altering the stereo's volume to get the tracks to an acceptable listening level. At first I hated the low volume CD's (typically a 3 track single) because they seemed so, well, 'amateurish'. Then, after a while, I noticed it was the super-loud CD's that pee'd me off because it was always a 'shock' when they blasted the speakers to bits!

Thinking back, it was a good window into the worlds of other amateur bands / artists. And I have to say, there were a lot of awful recordings, but no matter how bad they were, the good songs always stood out.