The rules of clipping

[inthecorner]

When I prepare a track for mastering, I make sure none of the channels individually clip.

I then look at my master fader and collectively drop all the channels of the song (as mixed) until I see no more clipping on the master output channel whatsoever.

Once there is no clipping present on the master channel I then proceed to add my mastering chain to the master channel/fader.

I start with EQ, then add the vintage warmer but I only add a little bit of bottom & high end, turn on the "FAT" setting and do I not touch the "GAIN" knob, I also make use of the "Brick-wall" limiting on this plug-in.

I then proceed to add iZotope's OZONE. I do multi-band compression but very slightly and then finally I boost the track using OZONES maximizer. I get a RMS volume that reaches just above -10dB.

My reason for describing my method is to ask whether or not the experts on this forum can tell me if it is ok if the processes that get added (EQ, Vintage warmer 2, Multi-band) are allowed to boost the overall volume of the song past the 0.0dB mark before the maximizer is finally added?....in other words when adding any mastering plug-in (alteration of sound and tone) to a pre-master (that has no clipping) is it ok if these processes (excluding the maximizer) allows the song to clip? Or must it not clip even when you add the processes to affect the tone etc. of the pre-master?

Your help and insight is much appreciated.

[MAzevedo]

Well, if you are clipping anywhere, you are adding distortion. Full stop, end of discussion. At very low levels of clipping, you might not be adding audible distortion, but any and all clipping is distortion. There are times when extreme final level is needed that a slight (or a not-so-slight) amount of clipping is needed to get you to a very, very loud final product, but if you are only shooting for -10dB you shouldn't need to clip anything.

[lucey]

Originally Posted by MAzevedo

Well, if you are clipping anywhere, you are adding distortion. Full stop, end of discussion. At very low levels of clipping, you might not be adding audible distortion, but any and all clipping is distortion. There are times when extreme final level is needed that a slight (or a not-so-slight) amount of clipping is needed to get you to a very, very loud final product, but if you are only shooting for -10dB you shouldn't need to clip anything.

+1 except the last sentence. Where no mastering compression is appropriate you can need clipping to get up to -10RMS.

[MAzevedo]

Well, if you want to have an RMS level of -10dBFS and not use compression, then yes you're going to have to clip it to get there for most material. If compression and/or limiting are on the table, then getting up to -10 isn't hard for a decent mix without clipping it.

[jamsmith]

Funny. Just posted on the subject in the humorous Mastering - Fully Explained thread. Not being (yet?) a professional mastering engineer I guess I don't yet have a handle on all the lingo, but in electronics, clipping is literally cutting of the top of a signal resulting in a waveform that resembles a square wave. Something you do with a zener diode.

I have seen people here use the term clipping to describe what audio engineers outside of mastering call hard limiting. (Here's the funny part: in radio, hard-limiting is achieved with a clipping circuit!)

Far be it from me to tell a group what to call things, but it is confusing to the rest of us to refer to anything that comes out of an audio compressor/limiter "clipping" unless the circuit is broken or you hit the input with a signal greater than it can take.

[lucey]

The rest of "us" is who exactly?

Clipping is clipping. Achieved with analog processing by overdriving the AD. ITB you might do it by bumping the master fader.

Limiting can be done with an analog process, or a digital one, or by clipping the AD.

[jamsmith]

Originally Posted by lucey

The rest of "us" is who exactly?

Those who are not "them".

[narcoman]

Originally Posted by jamsmith

Funny. Just posted on the subject in the humorous Mastering - Fully Explained thread. Not being (yet?) a professional mastering engineer I guess I don't yet have a handle on all the lingo, but in electronics, clipping is literally cutting of the top of a signal resulting in a waveform that resembles a square wave. Something you do with a zener diode.

I have seen people here use the term clipping to describe what audio engineers outside of mastering call hard limiting. (Here's the funny part: in radio, hard-limiting is achieved with a clipping circuit!)

Far be it from me to tell a group what to call things, but it is confusing to the rest of us to refer to anything that comes out of an audio compressor/limiter "clipping" unless the circuit is broken or you hit the input with a signal greater than it can take.

Ah - but you clip it on the way in - you drive an AD or hit inputs hard.

[peeder]

To answer the OP question, it all depends on the specific implementation of the DAW and plugins.

A floating point DAW has, in theory, 900db of headroom at every point on the board. So in theory would a floating point plugin. The only times such a thing would clip are when converting to fixed point, at the DAC or at the filesystem. But some programmers make their floating point plugins clip anyway, perhaps arguing they should behave the same as a fixed point plugin would. The better plugins have saturation models and they may soft-clip rather than just going square. apEQ is an example of a plugin where the programmers left it up to you to decide what the plugin should do (a good idea IMO).

Run a 500Hz sine wave through and see where things clip and where they don't, using an RTA or, after making sure you prevent clipping at the DAC, your ears.

I should do a video series on this too.

[inthecorner]

Interesting.

But doesn't the loudness maximized clip the audio anyway? I mean after Eq and multi-band compression you mos have to drop the threshold of the loudness maximizer to get the volume of the track up. When doing it you clearly see it's squashing the signal to get it to the desired RMS level.

[24-96 Mastering]

Originally Posted by inthecorner

Interesting.

So where've you been the last 2 years?

[inthecorner]

Apologies for the absence.....

[Cheebs Goat]

We need to back up here.

Originally Posted by inthecorner

When I prepare a track for mastering, I make sure none of the channels individually clip.

I then look at my master fader and collectively drop all the channels of the song (as mixed) until I see no more clipping on the master output channel whatsoever.

Once there is no clipping present on the master channel I then proceed to add my mastering chain to the master channel/fader.

If your mix is clipping, don't just yank stuff down and be done with it. Removing the clip will change the balance you set while mixing. Start out your mix at such a point that you will never run out of head room in the first place.

Which brings up the second question: How hot are you recording?

If you totally reset your mixing board/software and leave all of the faders at zero/all plugins off, the mix should not be close to clipping.

So...Maybe your problems start earlier than mastering?

[inthecorner]

Cheebs goat, thanks for the post.
Let me just say after dropping all the faders in the mix (i don't touch the master fader) I then turn up the monitors and listen to the mix and then alter it accordingly until it is a good mix again, so I don't just take the whole mix down and be done with it, I turn it down, turn up the monitors and mix it again, at which point there is more than enough headroom left on the master fader.

Let me also just mention that I mix with pre & post fader metering, making sure neither clip.

The second thing u mention makes no sense to me? If I turn off all the plugins on the audio and virtual instruments, stuff like compression Eq & fx sends, and set all of the channels (audio&virtual instruments) to 0db of course the master channel will clip, because it is the culmination of all the channels together. Don't really follow you? Maybe you could elaborate? I say this with humility. Thx.

[JeremyPaulCarrol]

I’d like to suggest that you get yourself a copy of ‘Mastering Audio - The Art and The Science’ 2nd edition by Bob Katz.

It is in my opinion an essential read for anyone approaching mastering. It will completely change your understanding of what you are doing and why you are doing it….

Cheers!

Jeremy.

[JonesH]

Why not just lower the master fader instead of individual channels? In floating point, this removes the clipping while still retaining the exact same balance.
In many instances though, people prefer to trim levels on all channels so as not to clip the master bus when everything is at unity gain. However, it's just a question of lowering the master fader gain, provided that the master channel inserts are post-fader (as in pro tools).

Please correct me if I'm wrong.

[echoRausch]

Originally Posted by JonesH

Why not just lower the master fader instead of individual channels? In floating point, this removes the clipping while still retaining the exact same balance.
In many instances though, people prefer to trim levels on all channels so as not to clip the master bus when everything is at unity gain. However, it's just a question of lowering the master fader gain, provided that the master channel inserts are post-fader (as in pro tools).

Please correct me if I'm wrong.

[alloysound]

Once you have clipped the master bus, lowering the master fader merely reduces the volume of a clipped signal. It's kinda like turning up the volume or drive on a guitar amp and then lowering the master volume pot... you're turning down the volume of a distorted signal.

If you find yourself with a mix that's clipping the master bus, then you can turn down each track... but you have to be REALLY careful to not upset your balance. It isn't as simple as just turning each channel down 3 db.

Say you have your drums bussed together and have a compressor plug-in. If you turn down each of the individual drum channels, you have changed the level feeding that compressor, and now your drums sound different and will sit in the mix differently.

You also have to think about FX sends and returns. If you are feeding an effect pre-fader, turning down that track will now give you a wetter balance. If you are feeding an effect post fader, then DON'T turn down the effect return.

If your tracks are hot to begin with, use a trim function or plugin to back the level down a bit BEFORE getting into mixing. If you try this after you're mix is too hot, you have to think about how that affects compressors and FX on EVERY channel.

And +1 to hitting A/D's a little lower when tracking, especially when you have many tracks. A/D converters can start to break up in an ugly way as the signal approaches 0dBFS. When you go to mix, plug-in's will have more headroom to handle changes to the signal, and the master bus will be less likely to clip with the accumulation of tracks that aren't too hot.

Leave it to the mastering phase to get the mix louder. ME's can make great masters from mixes that aren't slammed and clipped with way over-compressed channels in order to cram a bunch of tracks under 0dBFS.

[24-96 Mastering]

Originally Posted by alloysound

Once you have clipped the master bus, lowering the master fader merely reduces the volume of a clipped signal. It's kinda like turning up the volume or drive on a guitar amp and then lowering the master volume pot... you're turning down the volume of a distorted signal.

For many DAW applications today this isn't the case. Specifically, when the master bus operates in floating point (as JonesH wrote), it allows for positive levels. Within those applications, quite counter-intuitively, you can indeed simply lower the master fader to avoid clipping.

[csl]

Indeed. e.g. In Cubase, unless you're dealing with one of the rare fixed-point plugins (like some NI VSTs like Reaktor and Kontakt), nothing will clip unless your master output exceeds 0dBFS. If it does, then simply bring the master fader down; you're not losing anything except a sheared transient.

You could even export a 32-bit floating-point mix with plenty of apparent clips, then simply adjust the gain range during mastering down to something more workable. The only way such files could clip is during playback through your DACs.

[Cheebs Goat]

Originally Posted by inthecorner

The second thing u mention makes no sense to me? If I turn off all the plugins on the audio and virtual instruments, stuff like compression Eq & fx sends, and set all of the channels (audio&virtual instruments) to 0db of course the master channel will clip, because it is the culmination of all the channels together. Don't really follow you? Maybe you could elaborate? I say this with humility. Thx.

Right. This means you recorded too hot. The culmination of all of the channels together should not clip. Your recording peaks should be much lower. Don't be afraid to have tracks hitting -20 or lower.

I know the instinct is "then I'm not recording at full resolution". But think of it this way:
If you record with peaks around -4 and then mix with the fader down around -14, that extra resolution is "lost" to the fader move anyway. Not only that, but now you are stuck nudging faders down towards the bottom of the travel and tiptoeing around your aux sends so you don't clip a plug. Also the possibility exists that you could clip the master buss and have to pull everything back and re balance.

If you peaked at -18 in the first place, you just go ahead and mix. Your faders are in the "fat part" of their travel hanging out near zero where it is easier to mix, nothing you ever do will clip any plug or buss, and you don't have to think about a damn thing on the technical side.

The easiest way to do this: Record your initial tracks (drums or whatever) hovering around -18 or -12. Record each overdub as close as possible to a final mixed level in relation to the existing tracks with all of the faders set to zero. Try to make the recording itself properly balanced with headroom remaining.

[inthecorner]

Originally Posted by Cheebs Goat

Right. This means you recorded too hot. The culmination of all of the channels together should not clip. Your recording peaks should be much lower. Don't be afraid to have tracks hitting -20 or lower.

I know the instinct is "then I'm not recording at full resolution". But think of it this way:
If you record with peaks around -4 and then mix with the fader down around -14, that extra resolution is "lost" to the fader move anyway. Not only that, but now you are stuck nudging faders down towards the bottom of the travel and tiptoeing around your aux sends so you don't clip a plug. Also the possibility exists that you could clip the master buss and have to pull everything back and re balance.

If you peaked at -18 in the first place, you just go ahead and mix. Your faders are in the "fat part" of their travel hanging out near zero where it is easier to mix, nothing you ever do will clip any plug or buss, and you don't have to think about a damn thing on the technical side.

The easiest way to do this: Record your initial tracks (drums or whatever) hovering around -18 or -12. Record each overdub as close as possible to a final mixed level in relation to the existing tracks with all of the faders set to zero. Try to make the recording itself properly balanced with headroom remaining.

Thanks a lot very useful advice!

[inthecorner]

Originally Posted by Cheebs Goat

Right. This means you recorded too hot. The culmination of all of the channels together should not clip. Your recording peaks should be much lower. Don't be afraid to have tracks hitting -20 or lower.

I know the instinct is "then I'm not recording at full resolution". But think of it this way:
If you record with peaks around -4 and then mix with the fader down around -14, that extra resolution is "lost" to the fader move anyway. Not only that, but now you are stuck nudging faders down towards the bottom of the travel and tiptoeing around your aux sends so you don't clip a plug. Also the possibility exists that you could clip the master buss and have to pull everything back and re balance.

If you peaked at -18 in the first place, you just go ahead and mix. Your faders are in the "fat part" of their travel hanging out near zero where it is easier to mix, nothing you ever do will clip any plug or buss, and you don't have to think about a damn thing on the technical side.

The easiest way to do this: Record your initial tracks (drums or whatever) hovering around -18 or -12. Record each overdub as close as possible to a final mixed level in relation to the existing tracks with all of the faders set to zero. Try to make the recording itself properly balanced with headroom remaining.

What I just don't get is what about virtual stuff like battery 3, a lot of the virtual stuff clip internally anyway and if you put the faders of a couple (say 8 channels) of stylus rmx loops on zero your master fader will clip. So am I suppose to also make the internal level of the virtual stuff that soft? (-12db) I am applying the logic of this question to the experiment you made me do - putting all faders (including virtual instruments) to zero and bypassing fx (eq etc.) and then checking the master to see if it clips, thereby assessing if you are too hot to begin with. Or does the too hot test only apply to audio?

[steveschizoid]

Originally Posted by Cheebs Goat

I know the instinct is "then I'm not recording at full resolution". But think of it this way:
If you record with peaks around -4 and then mix with the fader down around -14, that extra resolution is "lost" to the fader move anyway. Not only that, but now you are stuck nudging faders down towards the bottom of the travel and tiptoeing around your aux sends so you don't clip a plug. Also the possibility exists that you could clip the master buss and have to pull everything back and re balance.

If you peaked at -18 in the first place, you just go ahead and mix. Your faders are in the "fat part" of their travel hanging out near zero where it is easier to mix, nothing you ever do will clip any plug or buss, and you don't have to think about a damn thing on the technical side.

.

Cheebs is right in that you really need to get away from the 16 bit mentality. With 24 bits, quantization distortion becomes pretty much a merely theoretical issue.

On the other hand, the impact of different preamp gain levels upon the sound is not theoretical. Generally the higher you go past nominal level, the more distortion and the lower the S/N ratio.

However, recording something that you intend to keep at a low level in the mix at an overly low level is also questionable practice in my opinion.

There's no reason to even think about keeping your fader near zero unless you're using a real console, and you don't lose resolution, headroom, or anything else when you pull your faders down in the virtual world.

The only thing you need to worry about is operating your analog gear at its nominal level (unless you want to add distortion) generally 0 dBvu, and not clipping your plugins ITB.

Record your background vocals in your preamp's happy zone, then pull down the faders in the DAW. Nothing is lost.

[Cheebs Goat]

Originally Posted by steveschizoid

There's no reason to even think about keeping your fader near zero unless you're using a real console, and you don't lose resolution, headroom, or anything else when you pull your faders down in the virtual world.

Right, pulling down a DAW fader does not perform any destructive act on the recorded audio, so it is not "lost" in that sense. But the audio that goes into the summing equation, the audio that will eventually be heard, DID lose resolution because of the drop in volume due to the fader. Yeah, it can be undone with a simple fader nudge. Still, if you don't want it loud in the mix, why record it loud in the mix?

And I understand pushing a preamp for effect. Most pre's have two knobs: a gain knob and an output knob. You can push the gain to distort and then bring it back down with the output before it hits your converters. At this point it's really just a workflow thing. Yeah, it might sound about the same if you don't attenuate the output and then pull down a fader in a DAW. I just find the work goes much faster and my brain is free of technical worries if I record at the final intended level.

[steveschizoid]

You don't lose resolution (in the sense that your signal-quantization noise ratio would change for the worse) by reducing your DAW fader volume. It's just math. Computers do that really well.

Whereas, if you record intentionally at say -40 dBfs/-24dBVU then you have lost "resolution" permanently.

In a 24 bit world however, we are likely quibbling about minutiae.

I'm not really a big fan of the "pushed" preamp effect. If they're happy at 0 dBvu, I'm happy too. I mainly mention it to appease the guys that hate rule based gain staging. (I wish Peeder was still around to argue!)

[Cheebs Goat]

Originally Posted by steveschizoid

In a 24 bit world however, we are likely quibbling about minutiae.

Indeed we are. Like I said, it is a workflow that works for me. Other methods are not incorrect.