Digital Clipping: So What?

  • Thread starter Thread starter DavidK
  • Start date Start date
DavidK

DavidK

New member
Yeah. So What? :confused:

Lemme explain:

I understand distortion, its fairly obvious to hear. If something goes above the 0.0 mark, is at bad if there is no distortion aurally? Is that possible? Am I just not hearing it?

I find that anything sustained above the zero mark will probably distort, but sometimes a fast "pop" will not. Is there any reason I should avoid this if it makes no difference sonically? Does this mean I am gay? :eek: Or just sloppy?

I have some stuff that does it. I dont intend on having it mastered, I do understand that you gotta leave room for mastering. I never clip on the individual tracks. What's up> Bad idea? Thanks.
 
DavidK said:
If something goes above the 0.0 mark, is at bad if there is no distortion aurally? Is that possible? Am I just not hearing it?
If you're working in 32-bit float format, going above 0 dBFS is not a problem as long as you bring the level down before it hits the D/A converters. If you're in an integer format, there's no such thing as "above" 0 dBFS. If you have more than a few consecutive samples hitting the top, you will probably be able to hear it. If you don't hear it, (or worse, if you like the sound!) I suppose it's OK.

Just be aware that someone with better monitors, or better ears, may be able to hear it.

Don
 
Don't forget about playback on consumer systems. If it goes above "0" it will distort. Sometimes even before that. Best to leave a little headroom in there.

I think one thing that people forget is how loud digital "0" is. When you are using your volume knob to control playback level you are using it as an attenuator. In other words, you are turning the volume down, not up. So when you turn the volume up you are really just turning it less down, if you get what I mean. Digital "0" is LOUD.

So there is plenty of room to leave some space up there. Like make the peaks no higher than -.5 to -.3, for example.
 
A small point, but I think it's digital overs that you're saying isn't a problem. Digital clipping is very audible, and very much a problem.
 
DonF said:
Just be aware that someone with better monitors, or better ears, may be able to hear it.

Don

My ears are as constant as the Northern star :rolleyes: :D

This is what I was wondering, thanks. My monitors are old and shot and I dont have a particularly good amp.

I dont really understand the first part. I work in Sonar3, I dont know the difference between 24 bit and 32bit floating whatchmacallit.

If you're in an integer format, there's no such thing as "above" 0 dBFS. If you have more than a few consecutive samples hitting the top, you will probably be able to hear it.

I dont understand this. Here is what I understand with my limited knowledge:

At zero, the clipping light goes on in Sonar. Certainly if I have something at +6 its gonna be a distorted mess. If I have some things at +2, it will usually be a mess too. If I have some short attacks that shoot up to +1-2 or so it seems to be fine. I have Tannoy Monitors ( Old and worn) and nice AKG cans. I can certainly hear it on the cans on sustained things, and I am listening to stuff in every way I have ( Car, Stereo, Walkman blah blah).

I guess what I am asking is if there is a technical reason to avoid this if I get CDs printed? Thanks a lot for the help. :cool: I do understand that when I export to a 16bit final, the top level is gonna be zero. I am just wondering if there is anything else I should be aware of ( like this showing up on better monitors).
 
You're right, the occasional transient clip is often neither audible or dangerous.

However, I'd like to turn one of your questions around. Instead of asking whther there's any reason to avoid such clipping, ask youself if there is any good reason not to avoid it?

I would offer the answer, "No, there is no good reason not to avoid clipping." In fact I can ennumerate a few good reasons to avoid it altogether.

1.) At what point does a "transient" clip become a "sustained" clip; where does one draw the line? There is no line. "When it becomes audible" is not a great answer because the immediate response would be "audible to whom?" Just because you can't hear it doen't mean it's not audible on all other systems to all other ears. Getting into the habit of avoiding clipping whenever possible aviods this potential issue altogether.

2.)We hear the occasional coment in here that "it's OK to throw some peaks into clipping; that's a small price to pay to be able to squeeze an extra 2 or 3 dB of RMS volume out of the track." My response to that is that if one needs to throw their recording into clipping in order to get that extra couple of measley dBs it means one of two things: either they don't yet know how to mix and master properly - because if they did they could get that volume without having to clip - or the content of the signal just doesn't need those extra dBs and the engineer is chasing a ghost RMS number. Learn to mix and master peoperly, and clipping will not be necessary. Conversely, resort to the clipping as a short cut to volume, and you'll never learn some better mixing and mastering techniques that will come in handy for other situations than just chasing RMS.

3.) As you mentioned yourself, if the stuff is being sent out for pro mastering, you not only don't want to send the ME something that clips, you'll want to send him something where there is headroom from him to work.

G.
 
You other guys snuck in :D
Are you talking about tracking, mixing or mastering?
Mixing.

I never track and clip. I dont intend to have this "mastered" by an outside source. I am talking mixing in Sonar and having total levels go above the zero mark. Sorry for my lack of technical jargon, I am a good musician but a terrible technician :p
 
SouthSIDE Glen said:
Getting into the habit of avoiding clipping whenever possible aviods this potential issue altogether.


3.) As you mentioned yourself, if the stuff is being sent out for pro mastering, you not only don't want to send the ME something that clips, you'll want to send him something where there is headroom from him to work.

G.

I was thinking that the first one might be a good lesson in not being sloppy. Here's the deal:

I have always been pretty meticulous about this. Recently, I have been doing very complex stuff. I THINK (?) I can hear the difference but I dont know and I guess there is no way to know unless I had 20 monitors.

AFAIK this is not being mastered. I am assuming that I want to get this as loud as humanly possible without clipping or distorting. Its those peaks I am mainly concerned with.

Thanks for all the good responses, this is very helpful info fellers :)
 
DavidK said:
I dont know the difference between 24 bit and 32bit floating whatchmacallit.
In 24-bit integer format, 0 dBFS is an absolute upper limit. In 32-bit float, 0 dBFS is just another point on the scale. BUT (and this is important) A/D and D/A converters work in 24-bit integer format. So it's impossible for the A/D converter to produce a signal over 0 dBFS, and it's impossible for the D/A converter to accurately reproduce a signal (generated within the DAW) that goes over 0 dBFS.

So, in a DAW that uses 32-bit float internally, you don't have to worry about clipping on a bus, for example. But you still have to be concerned about clipping anytime the A/D or D/A converters are in the signal path.

I agree with what Glen said, BTW. Leave yourself some headroom, and save the final gain adjustment for the mastering stage (even if it's a DIY mastering job).

Don
 
This "loud as humanly possible" thing is not quite that simple.

There is the actual volume level, which includes all the peaks and valleys. Then there is the "apparent" volume level, which is the overall level of the track that your ear *perceives*.

You can have very high peaks and yet still have a track that seems too soft. This is because the valleys between the peaks are long and low volume.

The way to address this is with a brickwall limiter. This limiter will bring up the lows and trim off the peaks, giving you more overall volume level, and higher "apparent" volume level from the perspective of the listener.

This peaking above zero is technically wrong and completely unnecessary. You run the risk of spending money on a run of CD's or on CDR media only to have your listeners complain about their music players are distorting. Then you'll have to go back and do it all over again.

If you want to take that risk, then go for it. But there is *no reason* to have individual tracks or mixed peaking above digital zero. Just turn it down the couple of dB and put a brickwall limiter on it.
 
I agree with what Glen said, BTW. Leave yourself some headroom, and save the final gain adjustment for the mastering stage (even if it's a DIY mastering job).

THIS is the part I dont understand :(

Overall I did a decent job of recording I assume, the thing sounds OK and the music is well done. I dont understand the mastering process, especially DIY. I know what mastering is in regards to levels etc. I dont know what I can do at home to "master" something.

I have Wavelab 4, which I dont know how to use very well ( its legit and I have the Manual of course). I mainly use it for editing lengths, making mp3s,burning CDs etc. I dont undersatand what I can do OVERALL to master this stuff that I cant do in SONAR. I will start learning about it, I am getting pretty serious and am about to complete the music aspect of a substantial CD in days.

THANKS GUYS!
 
In pro tools, at least, I've heard that the clip light goes on somewhat below 0dbfs. So the red light doesn't mean you HAVE clipped, it means your way too close for comfort and have probably clipped.

If you don't hear the clipping (and you trust your ears and monitors) then you don't have a problem. At least that has been my experience.

Its more art than science here. The science says "NEVER light the clip light!!" The art says "Well, that's a freakin' great take and I don't hear any distortion."

Listen to your music, don't look at it. If the clip LED lights up, listen CAREFULLY and back off the gain on that track for the rest of the takes. No need to flirt on that ragged edge, but also no need to discard a track that sounds good even with the little red lights.

-C
 
You guys are too cool for school :cool:

The way to address this is with a brickwall limiter. This limiter will bring up the lows and trim off the peaks, giving you more overall volume level, and higher "apparent" volume level from the perspective of the listener.

This is my next step/goal: learning about limiting. I have tried to learn fairly well about the recording process and now I am trying to get into the mastering. When I use a limiter do I use it in Wavelab or SONAR?

I have a LOT to learn. I'm frightened, hold me :(
 
DavidK said:
You guys are too cool for school :cool:



This is my next step/goal: learning about limiting. I have tried to learn fairly well about the recording process and now I am trying to get into the mastering. When I use a limiter do I use it in Wavelab or SONAR?

I have a LOT to learn. I'm frightened, hold me :(


'K DavidK, your music is sort of an interesting conundrum. Normally with classical music, you wouldn't dream of doing things like limiting or having transient overs. Or at least, you wouldn't tell ;)

But your music is sort of a hybrid, so you need to take a hybrid approach. Your music is still MUCH more dynamic than your average rock tune. If you are trying to narrow the dynamic range to make it more friendly when played next to rock tunes, you really need to start addressing that in the mix by bringing up the quiet parts and them turning them back down to the normal mix level when you've got the fortissimo goin' on.

Once you've done that, try setting a limiter on your master bus with about 1 dB of limiting. That will be fairly transparent for most of your tunes, and you'll get a bit of clip-free volume. Then try turning it up until you can hear the distortion on your loudest parts. Distortion from a limiter doesn't sound like clipping distortion, so listen critically. Once you've done that exercise, you'll have not only a good idea of how much limiting you can get away with, but also what elements of your mixes are causing the peaks, and how to deal with them in the mix so the limiter doesn't have to.

As for self-mastering, you can do it in Sonar or Wavelab, but Wavelab is designed to do it, so you might find it easier there. I think Wavelab is fairly easy to use; I can't hardly figure out sequencer software like Sonar :o
 
Excellent stuff MS. Really. :cool:

Yes, my weirdo stuff is a hybrid. I wouldnt expect a "classical" CD to be messed with but mine is more Vangelis than Beethoven.

Since I have you in my evil power :eek: :

I am sending my weirdo stuff to some small independent labels. I am gonna assume they dont master stuff. would you? :confused: They are used to violin Sonatas et al, I doubt they are gonna master it so I am thinking I have to do some kind of a reasonable facsimile of this on my own. Actually, one company already has most of it and is considering it already. :cool:

I am going to try your limiting advice and see what happens. I THINK I am in the ballpark as to general loudness, it compares OK to a commercial CD. There are just a few tracks that I want REALLY loud and I am not getting it so I will try limiting.
 
Going over 0dbfs on a channel in Sonar isn't clipping. Going over 0dbfs on your output buss is.

If you are going over 0db on your channels all the time, you are probably tracking too hot, but that is another topic.
 
Farview said:
Going over 0dbfs on a channel in Sonar isn't clipping. Going over 0dbfs on your output buss is.

If you are going over 0db on your channels all the time, you are probably tracking too hot, but that is another topic.

My levels are fine on the individual tracks, in the -6-3 range at the most. Its the combination of 150 tracks thats the problem :eek: I do some big stuff!

Yes, I am talking about the output buss ( I guess, I call it 'the thingy on the far right that says M-audio firewire on the bottom of it).

To put this into perspective:

Here is a track that IMHO is done. I know its not Kosher to judge by a lousy mp3, but I dont hear any distortion or clipping on the mp3 or the wav. Umm, does anybody else? If so I need to change some stuff.

link to mp3 of Firebird- final version

I would really appreciate it if one of you rapscallions could check it out. The first 20 seconds is all you need, Its as loud as the rest.
 
Whoa! Music with actual dynamics! What a novelty. :D

Sounds pretty good here at work, listening on cans. I did not notice any clips. I may take the cans off and put it up on the speakers to see if it causes a riot.

Don
 
DavidK said:
Here is a track that IMHO is done. I know its not Kosher to judge by a lousy mp3, but I dont hear any distortion or clipping on the mp3 or the wav. Umm, does anybody else? If so I need to change some stuff.
Dave,

As one of the resident rapscalions, I gave it a spin or three. First let me say, *very* nice work! You make Tomita sound like he is playing with rocks and animal skins. ;) (entirely different technology, I know, but still an amazing difference :))

As for the clipping. I can't say with 100% certainty that I am hearing that, but I will say that the high frequency punctuations in each bar at the beginning do have just a bit of a shrill or harsh edge, to my ears anyway. It's a temporary sound reminiscent of the first generation of commercial CDs when the converters had a "brittle" sound to them.

It's not a huge problem with your recording, but if I had to play devil's advocate, it's there and (IMHO) worth tweaking if you're going to be sending this to some labels.

From the MP3 I can't say for sure whether it's actual clipping causing what I hear, a conversion or dithering issue, or just a matter of dynamics or EQ in general, but even if you had not mentioned the clipping I would have thought those flourishes were just a bit sharp.

G.
 
Last edited:
Back
Top