[LAU] maximum input level, or normalization and dc offset correction?

[Ralf Mardorf]

On Fri, 2013-07-12 at 16:20 -0400, Ricardus Vincente wrote:
> On 07/12/2013 02:36 PM, Harry van Haaren wrote:
> 
> > <wizardofgosz at gmail.com <mailto:wizardofgosz at gmail.com>> wrote:
> >> Some people believe in this new gremlim called inter-sample distortion
> > or some such thing.
> > 
> > [Side note on gremlins]
> > Inter sample peaks: neither peak represents the highest value of the
> > wave: the peak exists between the samples. This is a mathematically
> > proven phenomena. The distance to keep from 0dB FS depends on the signal
> > (due to the inter-sample peaks depending on the signal).
> > 
> > I tend to stay away 3dB from 0dBFS, I think that suffices... -Harry
> 
>  I understand that they can exist, but for the waveform to be rendered
> by the D/A, 3dB over the sample values (in the peake between the
> samples) seems like an unlikely transient.  Further, if the D/A has
> sufficient headroom (modern good D/A should) I would think it's not
> really a problem.
> 
>  I still see plenty of mastering engineers normalizing their INCREDIBLY
> LOUD mixes to 0, and to -.1

And those mixes do sound good ;)? They mix music in a way that I can't
stand to listen it. You like to listen to this music?

>  So in short, I tend not worry about it.   :-)

In the past I made many masterings with DAT having a margin of just 0.5,
IOW -0.5 dBFS, when recording a live session without limiter even peaks
that were > 0 dBFS don't cause audible effects.

However, what do the meters show? Full-Scale Sine Wave, then the sine is
at 0 dBFS or Full-Scale Square Wave, then the max for the sine should be
at -3 dBFS. I really don't know what the meters do display.

The sound of DAT recordings with peaks at -6 dBFS or lower -12 dBFS etc.
don't sound less good, then recordings with peaks at -3 or 0 dBFS.

It's nonsense to take care about "optimized" leveling, there simply
should be headroom so that peaks won't reach 0 dBFS or -3 dBFS. As
somebody already pointed out, e.g. -20 dBFS, then peaks anyway might
reach 0 or -3 dBFS. For the mastering, when we know the peaks we don't
need that much headroom, but a peak won't reach always the same level,
sometimes it might be at -2.5 dBFS and sometimes at -3 dBFS, so minimal
headroom is needed. Even if > 0 dBFS sometimes isn't audible, sometimes
it does cause audible effects.

Regarding to noise, as Julien already pointed out, the instrument or
microphone does cause noise, you won't hear the noise of a good sound
card. Julien mentioned the DX7, yes, my DX7 does produce audible noise
too, but recording my Matrix-1000 is like recording a soft synth, there
isn't noise.

FWIW there is a reason that adding noise sometimes does improve the
sound quality http://en.wikipedia.org/wiki/Dither :

"Usage

Dither should be added to any low-amplitude or highly-periodic signal
before any quantization or re-quantization process, in order to
de-correlate the quantization noise from the input signal and to prevent
non-linear behavior (distortion); the lesser the bit depth, the greater
the dither must be. The result of the process still yields distortion,
but the distortion is of a random nature so the resulting noise is,
effectively, de-correlated from the intended signal. Any bit-reduction
process should add dither to the waveform before the reduction is
performed."