nominal input levels

[Fulcrum]

This potential asshat issue has been bothering me for a while now. Maybe it shouldn't, as I am still getting usable tracks done.

I understand that if I had a sound card capable of 24 bit/96KHz operation I could potentially get away with recording at levels approaching -20dB digital. I understand also that if my sound card were merely capable of 16 bit/48KHz operation, I might be a bit more concerned with getting the levels as hot as I could without clipping.

My particular sound card falls somewhere between. It'll do 48KHz, but only at 20 bit. I suppose I could track as though it were only a 16 bit box, but then I understand I run the risk of overloading my inputs a bit. And if I track as though I have a 24 bit box, I worry that that might not be quite enough level.

I still have my documentation somewhere. When I am looking at the specs in the appendix, what figure will tell me the level at which I ought to be recording? In the absence of that information, am I best off steering a path between the methods I would use for 16 and 24 bit?

[ckerian]

I wouldnt even give it much thought.

are you interfacing a console with it? a preamp? instrument level devices like bass, or a keyboard? are they -10 or +4?

what app you running? the meters may or may not be useful. Better safe than sorry and we know the more ya hit digital with the shittier it sounds so just play it safe but it really comes down to the sound card.

and it'll be a 96kbps mp3 anyways right?

whatever you used for the Thomas Jefferson tracks was great so dont sweat the s/n ratio thing.

[Pimp-X]

FWIW, I try and hit as close to 0db in my DAW as I can :)

And maybe it shows.

:lol:

[ckerian]

It comes down to how your gear is calibrated too. Is odbfsd -18 dbu on your console or is is -14? wheres your analog gear clip?

I guess we just use our ears in the end. I calibrated my mixing board a year or so ago based on the meters.

I never did calibrate my meters. I probably should do that.

But I decided not to worry about it. as long as I dont clip the outputs or inputs on the interface I have been good.

[Mixerpuppet]

Gain Staging is an excersise...

I just listen to what I hear rather than trying to figure out signal levels... from a musical point of view..

But Calibration should come first as your baseline configuration...IMO from a troubleshooting and maintenance stand point.

[Fulcrum]

These are good responses.. I wondered if I was making myself crazy over nothing.

are you interfacing a console with it? a preamp? instrument level devices like bass, or a keyboard? are they -10 or +4?

Yeah, I've got a 12 x 4 x 2 desk as my front end. I believe I'm running at +4-- that was what I set my Layla control panel to anyway. I'm mostly ITB anyway these days but when I do have to record one of my brick and mortar synths it always goes through the Carvin desk first.

what app you running? the meters may or may not be useful.

Sonar 6 PE. I like the meters-- between that, SPAN, and the VUs on my desk I imagine I get close. I was just curious about whether I could get it better... but I'm happy with what I'm doing so I'm not gonna worry about it any more.

whatever you used for the Thomas Jefferson tracks was great so dont sweat the s/n ratio thing.

Ha! I forgot I did that for ya.

[oudplayer]

I actually think it's a big deal, for a few reasons:


when you get used to setting input levels so that they are .03dB shy of red, you will likely track takes that go well into the red
your converters may distort before red appears on the screen
if you know a sound will be buried in the mix at -30dB you have no need for a 0dB level track and will gain nothing from it. You're more likely to add unwanted noise through bad gain staging.


I try to bring stuff in/track as close to the level I think the sounds will end up at in the mix, and feel that this technique overall has greatly improved both the quality of my in-the-box mixes as well as greatly quickening the speed of mixing sessions.

[nobby]

my brick and mortar synths

Holy Wendy Carlos, Batman!

Talk about old school!

[PRobb]

16 bit I like to live at the top of the meters to get all the dynamic range I can. 24 bit I take advantage of the extra dynamic range by living in the middle of meters to get extra headroom both in the ADs and in the DAW's mixer. If you're 20 bit you're somewhere in the middle. But remember that each extra bit doubles dynamic range so I would go more to the 24 bit approach.

[padboy]

16 bit I like to live at the top of the meters to get all the dynamic range I can. 24 bit I take advantage of the extra dynamic range by living in the middle of meters to get extra headroom both in the ADs and in the DAW's mixer. If you're 20 bit you're somewhere in the middle. But remember that each extra bit doubles dynamic range so I would go more to the 24 bit approach.

Ayuppers. I go 24 bit and sorta kinda push it towards the top. But I make sure my analog gain staging never gets into the red. The one that kills ya is doing percussion, and I get very conservative when I'm doing them.

There were these guys I worked with many 30x moons ago, they had a brandy new studio bought by their dad, and it had an MCI console and 2 Scotch M-79 decks. The one thing I ever heard from one of the brothers that actually made some sense was that...

"A snare has transients that exceed 22db over what the meter says"

Fucking parrot. He was full of shit, but it made me always think about it. And now that we have all these cool visual tools that can actually show you how badly you mangled the signal, nobody with ears to hear and eyes to see can only blame their own self for shitty sounding ingress of sounds.

What happens regarding the mix thing is a totally different equation, and is something that is more of an art.

That's just the way I feel about it. Feel free to tell me I'm blowin it out my ass.

pb

....pffffttt....

[AxeSlash]

What always gets me is that no two sets meters ever seem to have the same response time...so how the hell anyone is meant to judge whether a track is into the red when the meter takes about 3 years to reach the top anyway is beyond me. That point about the snare has some truth to it - if your meters are slow, they're gonna miss some of that transient at the start and you've then got a big misrepresentation of what's actually happening.

I like meters that are faster than an SR-71. At least you've got a fighting chance.

Thankfully my interface has an LED that goes red when the converter clips (but not a proper meter, damnit) so I can at least get that part of the system into the same state as the ballpark. Outside of that it's guesswork and ears mostly for me.

[philsaudio]

As far as meters go, if I really want to know what is happening (or more like what happened) I use the Elemental Audio Inspector plugin. Free and killer.

This whole thing about the 16 vs 24 bit dynamic range really cracks me up. No matter what the soundcard, the noise floor is going to be determined by the gear before the soundcard, like the preamp and or console etc. If you change to 24 bit and reduce the level your noise floor does not change unless you change the source.

That being said I try to keep everything 3- 6 dB lower than 0dBFSD in 16 bit on the way in. The DAW has plenty of gain to boost it from there.

24 bit? Sheesh it is gona mostly be played from a 128K MP3 anyhow.

[Bob Olhsson]

Never going higher than -6 or even -10 can improve sound quality dramatically because all but the most expensive converters have crappy underpowered analog stages. With traditional pro analog gear it's often good to error on the side of too hot but with pro-sumer analog and especially with pro-sumer digital I think it's much better sounding to error on the side of too low.

Given the choice, I would never record anything serious at 16 bits because it can't take nearly as much processing without the sound quality falling apart.

[Trazan]

It may be that I didn't read the thread thoroughly enough, but I didn't see it mentioned: You have 6dB dynamic range per bit, meaning 24 bit has an extra 48dB(!) over 16 bits. 20 bit has an extra 24dB, so you could peak at -23dBfs and still be "better" than a 16 bit card.

I'd just stay safe though. :Coolio:

[Bob Olhsson]

There really haven't been 16 bit cards in years!

The SMPTE, EBU and NAB did a bunch of research to determine ideal digital recording levels. What they came up with was 20 bits of resolution with an average recording level of -20 dBfs. using VU meters or -18 dBfs. using RMS metering.

Then along came the ADAT...

[Fulcrum]

Basically, I've just been careful not to peak over 0dBfs-- so my RMS levels have been 6-10dB or more below peak pretty consistently. I mean, I've been anal about it, stopping tracks and restarting them with the fader pulled back a little bit even if they were 95% done.

I guess my concern was whether I was overloading my input circuitry by recording that hot. It would seem that I'm not; I would think that because otherwise I would be hearing the sound breaking up.

I guess the other Layla users I've spoken to are correct, that it seems to be more forgiving of recording at relatively hot levels.

Once I get my next card, though, I will probably relax that and record at slightly lower levels. In the meantime, I'm not going to worry about it much.

Thanks all for your opinions and your science.