"proper" level of DB when recording..??

Well you want headroom. But at the same time you don't want a weak signal.
So... it's a tradeoff.

The weaker the signal... the less bits you are using!
But if you hit 0dB you can cause some nasty distortion.

I try to keep it between -3 and -10. As close as I can ot 0 without hitting it.

Also keep in mind instruments like distorted guitar, or compressed bass are not very dynamic. So they won't need as much headroom.
But drums, and vocals can be extremely dynamic. So be sure to leave them a lot of room to breathe.
 
tarnationsauce2 said:
As close as I can ot 0 without hitting it.
Click to expand...
This is a perfect example of bad engineering and myth spreading.

All of your equipment was designed to run at line level. 0dbVU.

Digital equipment is no different. Line level = -18dbFS (depending on how it is calabrated)

VU meters (the ones on analog equipment like preamps and mixers) are slow, they read the average signal level. There are transients that go well beyond what the meter is reading.

Peak meters (the ones in all digital equipment including DAWs) are very fast and read the peak signal level. The average signal level is well below what the meter reads.

Proper gain staging is to run your preamps, compressors, etc... at line level. When you feed that into a converter, your peak level will differ from track to track. That isn't important!! The important thing is that your average level is at line level. (and that the peaks don't excede 0dbFS, percussion is the exception to this rule because the transient to sustain ratio is very high)

Bit hording is a waste of time and causes more problems than it's worth (and more than I have time to type right now)
 
Farview said:
...and that the peaks don't excede 0dbFS...
Click to expand...
That's what I said. i don't know if yu interperated it differently.
I meant as close as possible to 0 PEAK.

Yes we know analog meters read average. But these days we have to go by peak. Your peaks should never go above 0 (for example or +/-32,768 samples 16-bit).

When I set gain, as a reality check, I run a session and check all the waveforms to be sure none of them are clipped from hitting hit 0.
 
The point that Farview is making is that you really don't ever want to get even *close* to -0dBFS. That's 18dB into the headroom of the front end. More bits = more resolution - That's true. But at the expense of distortion, lack of focus, noise, etc. - 24-bit digital wasn't introduced so you could record louder - It was introduced so you could record *quieter* at higher resolution.

You need to understand how *analog* works before digital will help you.

On top of that - Recording everything peaking near -0dBFS just so you can turn it all down 15 or 18dB so you can mix it... Does that really make sense to you?
 
tarnationsauce2 said:
That's what I said. i don't know if yu interperated it differently.
I meant as close as possible to 0 PEAK.
Click to expand...
This is the problem with what you said. If you read the rest of that sentence, you would have noticed that it was pertaining to percussive instruments and that it was the exception to the rule.

And, no, 0dbVU is just as important as it ever was. Other than making sure that you don't exceed the limit, they are pretty useless. Average volume is what you hear, not peak. When you try to make something louder, you are trying to raise the average volume. (and consequently use compressors and limiters to lower the ratio between the peak and average levels)

Re-read my post. There is a lot of information in there, you seem to have missed most of it.
 
Ok I think it's mainly Semantics getting in the way here.
What you said is exactly true. And so is the never cross 0 rule.
Though, I must be a "bit horder :D" for getting up to -10dB. Though, if It never crosses 0 then no real harm is done except that it may be louder (on average) than something like percussion that's average is very small, but peaks quite high.

In the old days of tracking in 16-bit and mixing in 16-bit, you seriously would loose a lot of resolution, probably gave me a bad habbit by todays standards.

So you guys think I should shoot for -18dB average?
 
What I took issue with was the "as high as you can" part of the statement. When things were 16bit, most converters were calibrated so that line level was -12db. Now line level seems to be about -18dbfs.

The problem with running your levels harder than line level is that you are running your mic preamp (mixer, compressor,etc...) beyond where it runs efficiently. The cheaper the equipment, the bigger the problem.

You are running your entire signal chain out of headroom.

As soon as you get it all into the computer, you have to turn all the faders down to keep from overloading the mix buss. You also run your plugins out of headroom as well. It just generally screws up the gain staging of the entire thing.

BTW, if you ever hit -6dbfs, you have used all your bits.
 
Uh, Jay, John... walk me through what you've said again, only use small little words for dummies like me, and be patient when I ask stupid questions.

I've always thought that as long as a signal didn't trip the clip signal in CE or AA, that the sound wouldn't distort, and as near as I can tell, that's been true for me - as long as the signal doesn't pass the 0 dB mark and trigger the clip meters, the sound is undistorted. And I've always thought that the clip level (0 dB) was way more important than the average level, so when I'm recording, I always adjust levels so that the peaks hit about -9 to -6 dB, so that if I get too excited and get louder during the tracking, the peaks won't clip. (I do this cuz I don't like using compressors much when I record these days.) Again, I use the meters in CE and AA to measure the strength of the signal going to disk.

Okay, I'm very interested in what you have to say about what I've written here. Very.
 
I've just re-read the whole thread and I think my question boils down to this: at what level does distortion and noise start kicking in when you're recording digitally, and at what level does the noisefloor start to become an issue?
 
Now I'm gonna quiz some individual bits that I don't get:

"Recording everything peaking near -0dBFS just so you can turn it all down 15 or 18dB so you can mix it... Does that really make sense to you?"

John, I don't see what the issue is if I have to turn every track down 5 to 15 dB in the mix. It's just a quick level adjustment, and it doesn't affect the sound quality, right?

"As soon as you get it all into the computer, you have to turn all the faders down to keep from overloading the mix buss. You also run your plugins out of headroom as well. It just generally screws up the gain staging of the entire thing."

Farview, same thing. I don't get it. What difference does it make if I reduce the level of a track in the mix? As for plugins like EQ or verb, they're not problematic unless I record a track really hot, in which case I have to turn it down before I run the plugin on the track, but again, so what? (It seems to me that having to turn a track down is less problematic than having to turn a quiet track up because of the noisefloor issue.)

Okay, again, I'm really looking forward to hearing this. I think I'm gonna learn something.
 
It's not the digital part that's the problem - It's everything before it.

John, I don't see what the issue is if I have to turn every track down 5 to 15 dB in the mix. It's just a quick level adjustment, and it doesn't affect the sound quality, right?
Click to expand...

By that point, it's too late. The signal is already compromised. You can "accurately" turn it down, but the noise is already there. The focus is already gone. The S/N is already screwed up.

Your gear - Preamps, EQ's, compressors, CONVERTERS (yes, even converters) are designed to run most efficiently at 0dBVU. If you drive the signal too hard, it will still be accurately recorded - With the added distortion and lack of focus and noise it now has being pushed well beyond 0dBVU.

Here's the simple experiment -

Record something you can do well over and over - Strumming a guitar and fingerpicking or something with your preamp hitting 0dBVU - the meat of the signal should be around -20dBFS and peaks wherever they may be (probably around -14 or -12 for very short instances).

Then record the same thing with the signal dancing around full scale - You're going to have to turn the the preamp up quite a bit to do it (which is what a lot of people are doing).

NORMALIZE both files - Yes, you heard right.

Adjust the louder file down to meet the apparent loudness of the quieter file so you can A/B them.

The difference between the two should be VERY apparent - The one recorded with the levels where the gear is designed to run will sound cleaner, clearer, more focused, less noisy - Everything about it will just sound better.

Also notice that the S/N on the "normal" file will be *BETTER* than the "hot" file.

It's like I keep saying - Don't take my word for it. This is the way the system was designed to work - and for good reason. It's nothing new. This is how it's done at every single (professional) studio I've ever been in. You don't set levels in digital. You set it *before* it's digital. Digital has that headroom "built-in" precisely for this purpose.
 
Okay. I'll try it. But you speak with such confidence that I suspect you're right. Damn. This is so important.
 
Dobo, it took me a bit to really understand what they were getting at too.
But now that i get it it all makes perfect sense.

I might try out Massive's experiment. But I'll change it a bit.
I'll use 1 mic (or source of some sort) running to 2 channels and then to 2 inputs to the sound card.
I'll set up the 2 channels like Massive said.
Then do everything else Massive said.

I'll listen to the resulting audio.
Then I'll capture s/n data.
Finally I'll invert one of the files and mix it with the other (bringing all common mode audio to -inf) to precisely analyze what is left, presumably the noise and garbage resulting from the over-gaining prior to digital capture.
 
I think John explained it pretty completely, but there are a couple things I would like to clarify.


When you turn your preamp up more than you need to you:
1. Turn up the self noise of the preamp

2. Use up the available headroom and run it at levels that are beyond where the preamp was designed to work efficiently. The distortion and S/N specs of the preamp were measured at 0dbVU, at levels above that, the distortion and noise numbers are much different.

3. The compressors,( EQ, converter, or what ever the next piece in the chain) input will have too much signal coming into it. This has the same effect on that piece of equipment that it does on the preamp. If you are using an EQ, you are adding gain to the signal which you will have to attenuate.

The entire idea of gain staging is:
1. The preamp brings the signal to line level.
2. Each piece of equipment in the chain should be recieving an expelling a line level signal.

The headroom is built into the system.


If you record your peak levels at -47dbfs, you still have more resolution than the CD your recording will end up on. Recording so your peaks end up around -12dbfs is not a big deal.
 
Xstatic said:
Personally, I do not blindly track as hot as I can without clipping for a different reason. Every converter is calibrated for it's own levels when comparing to analog. The average seems to be about -18. This means that on a digital system, -18 would correspond to the analog meter's 0. A lot of equipment out there (well most of it) has been designed for your average level to be about analog 0. I see a lot of people tracking to digital using a compressor to tame peaks and peak levels at -3 or so, with the average level being only a little below that thanks to the compression. Even if their average level were as low as -10, that means there analog preamps are hovering around +8, or maybe even +10 when you factor in the compression. For a lot of preamps, this does not leave a whole lot of room and often is well into the point where the preamp starts to change its character and gets noisier. People are right to assume that converters do a better job when you hit them harder and use more bits etc... etc... etc... However, how great is that difference in reality? As far as I am concerned, that difference is not nearly as big as the differnce in sound quality from a constant and consistent overworking of the preamps and gain staging of the analog circuitry. What good is having slightly better conversion when all it does it more accurately converts a signal that is not really the proper signal to begin with? I do agree that for certain tracks hitting a preamp hard is cool. But every track? No thanks. Especially with cheaper preamps (M audio, Mackie, Allen Heath etc...) when you push them past their sweet spot, the signal degredation is more abrupt and it falls apart pretty quick. With most preamps, staying in the reccomended range will often increase things like dynamic response, perceived depth of a track, minimal noise etc...

With analog, levels in a mix can not blindly be turned up. Every bit of analog circuitry makes some noise so tracking to quietly and then slamming it with gain in a mixdown can certainly cause issues with s/n ratios and just plain old backround noise. With digital, it's pretty easy. By digitally gaining up a few dB, all you are adding as far as noise to a track is the noise that was already there (interpreted by the converters in the A/D stage. I find tracking with proper levels does not cause me problems in a mixdown. I have no problems getting my unmastered mixes to the level they need to be. I also find that the tracks are a little easier to work with. Maybe thats just me, but the manufacturers semm to think the same as well
Click to expand...
I thought this might help.
 
Thanks Jay. I followed your link for the other thread over here and was thinking about copying my post to here.

It seems to me that the big problem here is that people are only thinking about their converters. What exactly is a converters job? It is to accurately convert your analog signal to digital. Everyone is so worried about the resolution loss in their converters that they forget all about the most important part of the signal chain, the beginning. People that track too hot are screwing up their signal before their converter. This is bad for 2 primary reasons....

1.) Your siganl has already lost a bunch of its voodou magic because you are making your equipment do stuff it was not intended for.

2.) Your converter is about to accurately convert the signal that you just finished removing a bunch of fidelity clarity and depth from. At this point, you may wish your converters were not so accurate;)
 
xstatic said:
Your converter is about to accurately convert the signal that you just finished removing a bunch of fidelity clarity and depth from. At this point, you may wish your converters were not so accurate;)
Click to expand...

Nice... I like that one. :eek:
 
I'll try John's test this weekend. In the meantime, I'm thinking this is the most significant bit of learning I've had in a long time. And all this in a lowly software forum on a largely unmoderated site! Okay, John, Farview, xstatic - here's a link to a forum that I'm promoting because it *is* moderated (the whole board, not just the Cool Edit/AA forum). Please consider checking it out:

http://www.recordingproject.com/bbs/viewforum.php?f=66
 
You must log in or register to reply here.
Back
Top