Old 05-24-2019, 07:19 PM   #1
talustalus
Human being with feelings
 
Join Date: Dec 2018
Posts: 431
Default Recording levels into preamp, the -18dBFS myth?

Hi

I have read from various sources that when recording into outboard preamp gear, you should aim to get the level around -18dBFS based on the VU metering system of yesteryear (where, -18dBFS = 0dBVU).

I wonder what people here's thoughts on this are?

I'm trying to record vocals and I have to turn up my mic preamp gain trim way high to get my singing level to be around -18dBFS on average. The downside is that with the trim set high, I can hear the noise floor and it's off-putting.

I don't see why I can't just record quieter and then raise the level of the audio after the fact? -- I mean, I know I can do this, but is there any engineering reason why I should not?

Thanks.
talustalus is offline   Reply With Quote
Old 05-24-2019, 07:57 PM   #2
Tomm
Human being with feelings
 
Join Date: Nov 2007
Posts: 1,265
Default

If you are getting better results recording low and then raising it later, then go for it.

The -18dbFS things is not a rule, just a good guideline to give folks enough headroom. It's to teach people to not go too hot into the recorder because 0dbFS will sound really horrible and there's no recovering from that. So play it a bit safer and aim for lower.

The only issue going very low, like peaking around -30dbFS, is you may start to notice a lack of resolution or precision in the audio, and you can introduce additional noise from the noise floor of your recording chain. For this reason 24-bit recording is a bit better off, because typically at that bit depth the converters are less noisey along with of course giving more resolution.

But again, if you found the best way to get clean audio with minimal noise, then that's obviously your best choice.
Tomm is offline   Reply With Quote
Old 05-24-2019, 08:03 PM   #3
serr
Human being with feelings
 
Join Date: Sep 2010
Posts: 8,473
Default

If you know ahead of time what the absolute highest level hitting your preamp will be, you can simply set a level to not hit 0 or go over.

If you don't know that ahead of time - because you are recording something new and aren't clairvoyant - shooting for -18 should keep you distortion free even if a few surprises come along.

That's the main point. An rms target level that should result in the loudest peaks not going over.

You seem to be having the opposite problem! So... no problem at all (with clipping peaks)!
Curious. You singing into an SM-7 or something?
serr is online now   Reply With Quote
Old 05-24-2019, 09:25 PM   #4
talustalus
Human being with feelings
 
Join Date: Dec 2018
Posts: 431
Default

Quote:
Originally Posted by serr View Post
If you know ahead of time what the absolute highest level hitting your preamp will be, you can simply set a level to not hit 0 or go over.

If you don't know that ahead of time - because you are recording something new and aren't clairvoyant - shooting for -18 should keep you distortion free even if a few surprises come along.

That's the main point. An rms target level that should result in the loudest peaks not going over.

You seem to be having the opposite problem! So... no problem at all (with clipping peaks)!
Curious. You singing into an SM-7 or something?
Thx. Yeah, I guess another part to my question is -- how much noise floor being audible is acceptable? Maybe a little is normal, because I'm sure it won't stick out in a mix (it's at like -66dBFS), and could be treated in mixing anyway.

I am using an SM7B in this example.
talustalus is offline   Reply With Quote
Old 05-24-2019, 09:45 PM   #5
serr
Human being with feelings
 
Join Date: Sep 2010
Posts: 8,473
Default

Yeah, the SM7 has a low output. People with loud obnoxious voices that can pummel other mics like it.

If you were singing all quiet and delicate like, maybe noise could creep into the mix. But you're probably fine.

24 bit makes lots of things easy yes.
You could turn a source down 96db and still have 8 bits of resolution left!
Makes it easy to record and not worry about getting digitally grainy at low levels. As the consumer format, it delivers full fidelity and nuance no matter how dynamic your final master is.
serr is online now   Reply With Quote
Old 05-24-2019, 09:23 PM   #6
talustalus
Human being with feelings
 
Join Date: Dec 2018
Posts: 431
Default

Quote:
Originally Posted by Tomm View Post
If you are getting better results recording low and then raising it later, then go for it.

The -18dbFS things is not a rule, just a good guideline to give folks enough headroom. It's to teach people to not go too hot into the recorder because 0dbFS will sound really horrible and there's no recovering from that. So play it a bit safer and aim for lower.

The only issue going very low, like peaking around -30dbFS, is you may start to notice a lack of resolution or precision in the audio, and you can introduce additional noise from the noise floor of your recording chain. For this reason 24-bit recording is a bit better off, because typically at that bit depth the converters are less noisey along with of course giving more resolution.

But again, if you found the best way to get clean audio with minimal noise, then that's obviously your best choice.
Thx. Are you saying 24-bit recording maintains resolution even at low input signal recording?
talustalus is offline   Reply With Quote
Old 05-24-2019, 09:57 PM   #7
Tomm
Human being with feelings
 
Join Date: Nov 2007
Posts: 1,265
Default

Quote:
Originally Posted by talustalus View Post
Thx. Are you saying 24-bit recording maintains resolution even at low input signal recording?
it maintains MORE resolution, sure



Imagine recording a sine wave at a low level in both 16 and 24-bit. the 24-bit version will be more accurate because there are more "steps" to represent the shape

Quote:
how much noise floor being audible is acceptable
That's up to you, but what you should be concerned with is noise-to-signal ratio.



if you move the example on the right up or down (raise and lower the gain) it won't really make a difference because the signal-to-ratio will be the same, you would just be raising and lowering both the noise and signal, and when mixing it'll end up being about the same anyways.

what isn't shown in this graphic is there are other things that contribute to the noise, like that preamp maybe gets noisier at higher gain settings making the signal-to-noise ratio lessened. so you can't just use numbers and meters to figure this stuff out, you gotta experiment.

The SM7B needs a lot of gain on most voices, so the preamp needs to be pretty high quality. or try to sing louder or closer to it to raise the signal-to-noise ratio.
Tomm is offline   Reply With Quote
Old 05-24-2019, 10:31 PM   #8
mschnell
Human being with feelings
 
mschnell's Avatar
 
Join Date: Jun 2013
Location: Krefeld, Germany
Posts: 8,135
Default

-18 db means loosing 3 Bits, which is no problem with a 24 dB A/D converter. (Regaring that the resolution of a CD is 16 bits, and analogue Studio tape provides an equivalent of something like 14 Bits.)

hitting the digital limit at 0dB is a catastrophe..

Hence -18 dB is not a myth, but a compromise that seems useful in many situations.

Of course using an analogue compressor before the A/D converter allows for reducing the digital Headroom appropriately.

-Michael
mschnell is offline   Reply With Quote
Old 05-25-2019, 12:08 AM   #9
ChristopherT
Human being with feelings
 
Join Date: Apr 2017
Location: South
Posts: 575
Default

get a cloud lifter - or another brand.
This will get you more gain from the mic into your pre amp and will help.

Most people don't realize that the majority of budget gear has 10cents worth of parts in the mic pre - therefore they sound crappy when gained up.

If you are really into the sound of the (very low level) SM7B - then best to invest in a good preamp to get plenty of clean gain, and then after the mic preamp - go line in to your A to D convertor.

Found this on 1st search for cloudlifter!

https://www.youtube.com/watch?v=dfDPD3_3128

There is no -18dbFS myth - you have a very low level mic, and you also must have a mic pre that is not suitable for low level mics.

Last edited by ChristopherT; 05-25-2019 at 12:19 AM.
ChristopherT is offline   Reply With Quote
Old 05-25-2019, 06:35 PM   #10
ashcat_lt
Human being with feelings
 
Join Date: Dec 2012
Posts: 4,340
Default

Quote:
Originally Posted by Tomm View Post
it maintains MORE resolution, sure
This is utter bullshit.

-18dbFS is not really a myth, but it's not necessarily true that 0dbVU (= +4dbu) = -18dbFS for any given interface. You have to read and decipher your specs, and sometimes that means doing math.

The right thing for the OP to do is figure out where on the interface gain knob they get the best S/N ratio. Turning it up to where your signal averages -18dbFS gives you a certain measurable noise floor. Does averaging -30dbFS put the noise floor further below your signal? If you normalize them to the same average, which one has less noise. That's all that matters. Period.
ashcat_lt is online now   Reply With Quote
Old 05-25-2019, 07:34 PM   #11
Tomm
Human being with feelings
 
Join Date: Nov 2007
Posts: 1,265
Default

Quote:
Originally Posted by ashcat_lt View Post
This is utter bullshit.
what's bullshit? I think we're all in agreement that S/N is the most important thing, 24-bit just gives you more room to place your recording level as it offers more resolution and thus a lower theoretical noise floor than 16-bit
Tomm is offline   Reply With Quote
Old 05-25-2019, 07:45 PM   #12
talustalus
Human being with feelings
 
Join Date: Dec 2018
Posts: 431
Default

Quote:
Originally Posted by ashcat_lt View Post

Does averaging -30dbFS put the noise floor further below your signal? If you normalize them to the same average, which one has less noise. That's all that matters. Period.

This is a good test. I will do that.
talustalus is offline   Reply With Quote
Old 05-26-2019, 03:43 PM   #13
martifingers
Human being with feelings
 
Join Date: May 2011
Posts: 1,787
Default

Quote:
Originally Posted by Tomm View Post
Imagine recording a sine wave at a low level in both 16 and 24-bit. the 24-bit version will be more accurate because there are more "steps" to represent the shape

Is this correct? I thought that all 24 bit gave you was a lower noise floor. And aren't stair steps misleading?
https://www.theverge.com/2015/11/6/9...-res-explainer
martifingers is offline   Reply With Quote
Old 05-26-2019, 06:41 PM   #14
Tomm
Human being with feelings
 
Join Date: Nov 2007
Posts: 1,265
Default

yea it's correct, 16-bit signed is −32,768 through 32,767 while 24-bit signed is −8,388,608 to 8,388,607

"steps" is not the best word, "values" is better... and I wouldn't say "stair steps"... I did come close but yea it's a "bit" misleading

love that video btw, saw it years ago so it was a nice refresher. noise floor improves as he clearly demonstrates, but so does that quantization error (or jiggly movement in the waveform) of course you can't hear it in most cases, but he doesn't go into how that effect can become exaggerated at lower gain settings... there's probably a better video that demonstrates that out there
Tomm is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 12:24 PM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.