Gain Staging

[wolfskin]

Hey all,

Here's a question regarding the 002r gain structure. Per the manual:

"All Digi 002r audio inputs and outputs are
set for 14dB of headroom below 0dB, or full code."

Now, I take this to mean that (when recording at 24-bit) all input signals which exceed -14dB on the meters are utilizing the full 24-bit depth. Therefore, all signal between -14dB and 0dB are in the 'optimal' recording window.

Of course, optimal 'analog' recording technique (in general) was to record at the highest levels without clipping to saturate the magnetic tape.

Therefore, would it be accurate to say that within Pro Tools, recording at say -5dB (static) and keeping all peakes under 0dB and over -14dB would have no advantage (in sound quality) over recording at say -9dB (static) provided the peaks are kept within the same 0dB to -14dB window?

Any thoughts / insight would be appreciated!

Eric

[rockrev]

Recording at an average of -18 would sound just as clear as recording at an average of -6. When recording 24-bit, you don't even need to be "in the yellow."

rockrev

[wolfskin]

Thanks Rev,

However, is it true that we wouldn't be utilizing the full 24-bit depth
when recording at -18db? Again, must the signals exceed -14dB to use 'full code'?

[Chris Cavell]

Thanks Rev,

However, is it true that we wouldn't be utilizing the full 24-bit depth
when recording at -18db? Again, must the signals exceed -14dB to use 'full code'?

You're not quite grasping the full concept. You'll NEVER be able to use all 24 bits...no matter what level you track at. Here's why:

24 bits = a dynamic range of 144dBfs, each bit accounting for 6dB of that dynamic range. In order for ALL of the bits to contain musical information, you would need to capture a source where the dynamic range is more than 22 bits (or 132dB), and calibrate it so that the peak signal is greater than -6dBfs and the noise floor less than -138dBfs.

For all intents and purposes, that's impossible.

In most facilities and live venues, the ambient noise floor alone restricts you to less than 20bits of dynamic range. In these most common of situations, you could track with peaks in the lower 20dB's (or even much lower in some scenarios) and still capure all the musical information that you would capture if you tracked with peaks hitting near 0.

In other words, get a signal, and stop worrying about how hot it is...it doesn't matter.

[wolfskin]

Thanks Chris for your very insightful response.

This question really just came about in a discussion with a client last week. I guess the idea I was trying to convey is that within the digital domain, it's much more important to limit the peaks below 0dB to avoid clipping, whereas in the old analog world, a hot signal was key to eliminate / minimize noise (while occasionaly bouncing 'in the red' was acceptable, and sometimes even desireable).

This naturally led me to consider exactly what information the meters within PT are relaying to me, and what advantage, if any, is inherent with a 'hotter' signal. Hence, when I found the "...-14dB of headroom below 0dB..." blurb in the manual, it became clear to me that a 'cooler' signal level may in fact be more practical (and sound just as good).

I guess it's just never occured to me just how big of a window we have for a signal in the PT world...

Thanks again...

[Skyflash]

Check out this thread. I read this a couple of weeks ago and it was a major revalation
for me. There are links listed here to PDF's and forum discussions by some of the top
people in the industry.

http://duc.digidesign.com/showthread...sb=5&o=31&vc=1

I've discovered that most of my audio files recorded over the last 4 years have been too hot.
I'm going to be re-recording a lot of my stuff over the next few weeks.

[crazyfoo']

1 bit corresponts to 6db of dynamic range (roughly). Therfore every time you go 6db below the clip point, you are not using one of the LSBs.

But,,, I can pretty much guarantee that you convertor wont be able to hack as much dynamic range as the theoretical range of a 24 bit word can. NO convertor can have 144db of dynamic range. The dynamic range of our ears is only 120db anyway so it'd be pointless!



Im more of a live sound guy than a studio guy so I just gain my signals up so that they are the level I want in the mix with my fader at 0. only if this is seriously quiet will I gain it up more so that I can turn down the fader, but this is more because Im anal about digital stuff than because it really does make a sound difference!

[EGS]

Hi Wolfskin,

I love your scientific question! Here's some thoughts that steer away from science and move more toward art...

First of all, forget analog tape gain staging. It's a different ballgame when it comes to digital bit depth. I think Digi has done us all a favor by providing a lot of built-in headroom. It reminds me of the type of headroom that Dolby SR gave us in the good ol' 2" tape days.

If you set your tracking levels so as to not hit the red peak light, you are in great shape. While this sounds like a dumbed-down answer, I think it makes sense. If you set tracking levels this way, you'll be able to focus on other tracking issues that affect the sound quality MUCH more than bit depth. Mic selection, mic placement, mic pads, micpre selection, micpre in/out gain staging, etc. Find the level that does hit the peak light, then back it off a few db. If the performer is capable of lots of dynamics, then back it down several db, and be ready to ride the micpre input gain during takes.

Mastering, on the other hand, is much more about the science of achieving full-mode, maximum bit-depth recordings. For that, I recommend conservative input levels, and final output levels .3db below full digital 0. If you regularly max it out to full 0, you're risking digital overs that can distort DACs on players, and/or cause problems with the (seemingly inevitable) MP3 conversion process. There are some great DUC forum discussions on the topic of mastering, but I believe your original post was referring to tracking levels.

Happy New Year!!!