29 Mar
2007
29 Mar
'07
8:54 p.m.
Phil Mendelsohn wrote:
Message: 5
Date: Thu, 29 Mar 2007 23:38:30 +0200
From: Dragan Noveski <perodog(a)gmx.net>
Subject: Re: [linux-audio-user] Peak Level Monitor/Limiting
i think, recording at low level, you are kind of "loosing disc space".
lets say, if you are recording at 24bit, on low level the signal
information will take only 12 bit (example) and the rest of 12 bit will
be only empty data information.
don't think that one can repair that with some gain plugin or
normalization...
If you think about this for a couple of seconds, you may
decide to reformulate your opinion. In a low-level signal,
you say the "top" (Most Significant) bits will be zero. In
a high-level signal, though, the bottom 12-bits are all ones.
What is the difference?
In the computer, a signal is a signal, regardless of the
value at which it is represented. It is only when you try and
look at the distance from that signal to either the noise floor
or the max, or deal with non-linear perceptual effects (equal
loudness curves, etc.) that level may come into play.
It's not the absolute level that matters, it's the dynamic
range.
Cheers,
Phil M
still not agreed.
if you use 24bit you can 'describe' the signal much more exactly, as if
you use 12bit.
try to convert any soundfile into 8bit and listen to it!
so if you record at 24bit with low level, perhaps you ll ending up
'describing' the noise of you hardware, but not a signal.
and recording at low level means less dynamic, that's right.
cheers,
doc