Question about distortion while mixing and recording digital

[ChrisP]

When I load professionally recorded songs into my digital software, I always see the signal hit the red, hence there should be distortion right? I don't hear any. My ears are shit? Or is it not a given that a track will distort even if it is above 0 DB in a digital median. Does the 0 db apply only to when the track is recorded and there is more slack when it comes to mixing in digital or is it because the tracks where recorded in Analogue and they have more of a lenience for overall volume or loudness?


If I record my tracks digitally and the individual tracks are all under the clipping level of 0 dbs, does it mean my final mixdown will distort if the signal passes 0 dbs while I'm playing back the tracks collectively?

The songs I uploaded into my daw don't really sound distorted so what is going on?

I have seen multiple sources claim that tracks should not be recorded over -16 db anyway. Is this digital signals or analogue? Is -16 db in digital the same as -16 db in analogue?

But I just want to know how I can tell if it my songs will distort or not. Do I have to check the RMS levels? Can I mix, keep everything under 0 DB in digital, then pass it through a limiter? I heard you can use limiting to add an extra 7 dbs to your mix without ruining the audible dynamics of your song. Can I push it into the red without any artifacts/distortion being heard? If so, what is my margins?

 

[Steenamaroo]

Hi Chris,
It's not an absolute given that peaking your DAW meters will result in audible distortion.
It might be interesting for you to add a plugin with gain and play with intentionally pushing a track into clipping just so see/hear when there are noticeable effects.

None the less, it's not something to take for granted or aim for.

When people tell you to track at -16 or any other number, they really mean that you should leave plenty of room, for a few reasons.
One is that in the olds days, when analog signal path noise and tape noise was an issue, it was beneficial to track hot to keep the signal to noise ratio wide.
These days that's no longer an issue so old, or inherited, habits aren't necessary any more.

The other is that the master output is the sum of all other tracks. If you record everything as hot as possible then the more tracks you add the more you're going to be constantly rebalancing,
pulling everything back because you keep hitting the master too hard.

I can't really tell you much about the various scales, I'm afraid, but sticking with the super simple rule of just don't track hot for the sake of it will almost always serve you well.

RMS is an average. It's not really useful or relevant when talking about clipping, but it will tell you something about how loud your track sounds.
Perceived volume is very different from peak volume.
RMS is shaped by altering the dynamic range of the master, or the tracks which contribute. The less dynamic range, the higher you can push the whole thing, and the louder your track is going to sound.
Again, not saying that's a good thing or something to aim for, necessarily. It is what it is.

Hopefully someone will come along and talk about 32 bit FP. I don't know the ins and outs but I'm fairly sure it's possible to mix with clipping and not hear it until you bounce the thing down to 16/24 bit in some daws.


A rule of thumb is that a well mixed session, with perceived volume in mind, is always going to carry better than a weak mix which has been heavily compresded or pushed hard for volume at the final stages.

 

[bouldersoundguy]

There are two kinds of distortion possible with high digital levels. One is the actual sound of the clipped digital waveform. That can be quite subtle until you really clip it hard.

The other kind of distortion depends on the ADC. Some (older?) ADCs don't handle intersample peaks well. Those can add distortion to the signal. I suspect there aren't that many devices with this kind of ADC left in the world.

 

[snow lizard]

Is -16 db in digital the same as -16 db in analogue?

No.

There's a standard level for audio devices called line level. Two of them actually. Line level for consumer equipment like home stereo stuff is -10dBV. Line level for pro audio stuff like mics and PA systems is +4dBu. The dBV and dBu scales are not the same, so while these are standards that have been in place forever, and are measuarable and fixed, the difference between them is not 14 dB. (going from memory I think it's something like 12, not sure). These numbers are intended to be used with RMS values. For home recording purposes with a decent interface that has mic preamps the level we're interested in is the +4 dBu standard.

Long time ago with analog tape recording, people were trying to get relevant level information from the meters. Engineers came up with the VU meter which has an intentional 300 ms delay to its reaction time. The delay means that VU meters can't show you peak level information, but they are easy on the eyes and useful for setting average or RMS levels. There is no standard for where to calibrate a VU meter, that was intended to be up to the user. The way people started using them was pretty much always the same, though - so to keep things simple everybody pretty much calibrated to 0 dBVU = +4 dBu.

So at line level, the system is designed to have headroom above the reference level. How much headroom depends on the type and quality of the circuit. Any specific device might have say 20 dB headroom, give or take. So your signal would have to peak that far above the 0 VU reference level before overdriving the circuit and distorting.

With digital recording, the scale is called dBfs or "decibels full scale". Again, it's not a rigidly defined standard, so where "line level" is depends on where the digital audio converters are calibrated to. With dBfs, the 0 mark is the absolute maximum, after which there is nowhere to store the audio information. When the clip light comes on in a digital system, it means that something like 2 or 3 consecutive samples have gone over 0 dBfs. It takes a very small sliver of a second to do that. Sustained digital clipping sounds terrible, but if it's only the occasional transient peak from a cymbal hit or something, there isn't enough time involved for our ears to register distortion so even though the meters say it clipped, it sounds clean.

Again, where "line level" is in dBfs depends on where the converters are calibrated to. Some of the expensive, pro level converters are adjustable by the user, while most of the inexpensive ones used for home recording are not. You can check the manual to see what it says about your individual device, but line level is generally anywhere between -15 and -22 dBfs. -18 dBfs is a common average that many people work with. This means 0 dBVU = -18 dBfs or whatever. So again, there is headroom above the average reference level to keep your signal free from distortion and allow some room to capture the peaky, transient stuff. This is very important when setting your levels for tracking. If you run your levels too hot when tracking, you can easily make the preamp or converter distort. This distortion will become part of your recorded signal and there's not much you can do about it. By keeping your levels down around line level or even slightly under, you're allowing the signal to stay clean, with lots of headroom. It improves the sound quality over something that's been tracked too hot. Also digital is not like tape, so there isn't really a sweet spot to aim for to maximize signal to niose ratio. A signal recorded a bit too low won't really suffer any consequence.

Once the signal is in the digital domain, it's possible to crank it up without adding distortion. The mastered level of audio CDs has been getting louder and louder over time because of something called the "loudness wars". There's so much compression and limiting going on that it squeezes a lot of the life and dynamics out of the music, but you can get much higher RMS levels without a whole lot of - or any - digital clipping.

If you want to record and mix your tracks and then make them loud with a limiter, that should be the very last thing that happens to process the audio. The cleaner your tracks are to begin with, the more punishment they can take at that stage.

Can I push it into the red without any artifacts/distortion being heard? If so, what is my margins?

In analog? Sort of, yeah. Sometimes. It depends. Digital? Not really, no. Again, if it's only a stray transient peak that triggered the clip light and it sounds okay, you should be fine, but that should only ever be possible at the end of the process. Properly recorded and mixed levels should have enough headroom that you're not even close to 0 dBfs.