naturerecordists
[Top] [All Lists]

Re: EIN values and bits

Subject: Re: EIN values and bits
From: "Dan Dugan" dandugan_1999
Date: Tue Dec 4, 2012 9:33 am ((PST))
>>> I use 24-bit recording because it gives you extra headroom to work with.
>> You can
>>> record low for safety in the field and boost it in post without any loss
>> of quality.
>>> Now that storage is larger and cheaper, the larger file size doesn't
>> matter much.
> 
> I've never understood this. When you boost the level in post, why doesn't
> that raise the noise as well as the signal?

Your intuition is correct, raising the level in post raises both the signal and 
the noise.

> Why isn't it better to aim for
> the maximum possible signal level initially in order to end up with the
> maximum s/n at the end?

That was true when the major source of noise was the recording medium. But that 
is no longer true. The recording medium, digital code, now has a much wider 
dynamic range than the signals we are recording.

A thought experiment. Say in a natural environment a loud bird that lands near 
my microphones makes a sound measuring 85 dBSPL. The natural background noise 
in this location is 35 dBSPL. The dynamic range of this soundscape is 50 dB.

If I make a "hot" recording the way we used to with tape, I'd put the bird at, 
say, -5 dBFS. The ambient noise would then be recorded at a level of -55 dBFS.

But the noise level of the recording medium, 24-bit digital pulse code 
modulation (PCM), is around -138 dBFS. That means we have around 83 dB of 
unused range at the bottom of the recording.

So say I back off 15 dB on my recording level and record the bird at -20 dBFS. 
Now the natural ambience will be recorded at -70 dBFS. 

In post, I increase the gain 15 dB digitally. That raises the bird up to -5 
dBFS and the ambience to -50 dBFS. It also raises the noise of the PCM medium 
from -138 dBFS to -123 dBFS. But in the presence of -50 dB ambience that makes 
no difference at all.

QED. Does that help?

-Dan

p.s. For simplicity I'm ignoring the difference between peak and average 
metering. 

dBSPL means sound pressure level. dBFS means signal level referenced to the 
maximum or clipping level of the digital code. 

I'm also ignoring how the recording level is adjusted, analog before the A/D 
conversion or digitally after. Another topic.








<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU