The Logic Off Topic list

Like to end this argument once and for all. Is using the gain function
in the audio window better than normalizing your individual audio
tracks to bring the levels up to get the best quality recording to mix
with?

Drummerboy

>Like to end this argument once and for all. Is using the gain function
>in the audio window better than normalizing your individual audio
>tracks to bring the levels up to get the best quality recording to mix
>with?
>
Normalization and Gain are similar functions.  Normalization just 
maximizes the gain, whereas gain enables you to make incremental 
adjustments (both increases and decreases).

However, John Pitcairn posted an interesting comment on using Logic's 
normalization function (which I assume also applies to the Gain 
function).  Here is a snippet of that post:

>The reason normalization is frowned upon is that it's generally
>performed at a fixed bit-depth as a destructive process - this will
>permanently reduce the sound quality of the audio slightly, due to
>resampling quantization error. Emagic do not specify the processing
>bit depth for the Sample Editor, but it may in fact be done at the bit
>depth of the original file (16- or 24-bit fixed point), which is
>definitely a bad thing.
>
>In Logic, it is far preferable to increase level by boosting the fader
>level and/or inserting Gainer plugins, which ensures that the increase
>in level is done both non-destructively and at a much more accurate
>32-bit floating point resolution. The result when added up across many
> tracks is _clearly_ superior to normalizing all the audio.
>
Given the above comment, I would assume the best way to maximize your 
signal-to-noise ratio is get the levels set correctly when you record 
the sound.  Otherwise, gain, normalization, or using the Gainer plugin 
will raise both signal and noise. 

Kind regards,
Dave

On a fine day, 12-03-2004, drummerboy5142 wrote:

>Like to end this argument once and for all. Is using the gain function
>in the audio window better than normalizing your individual audio
>tracks to bring the levels up to get the best quality recording to mix
>with?

The best way (except for recording at decent levels :) is to use the 
Gainer plug-in.  Since it's a plug-in, it operates in Logic's native 
32-bit space.  All destructive edits, such as normalizing, work on a 
24-bit level and introduce quantization errors (noise).

amgshaffer quoted John Pitcairn and wrote:
>  >In Logic, it is far preferable to increase level by boosting the fader
>>level and/or inserting Gainer plugins, which ensures that the increase
>>in level is done both non-destructively and at a much more accurate
>>32-bit floating point resolution. The result when added up across many
>  > tracks is _clearly_ superior to normalizing all the audio.
>
>Given the above comment, I would assume the best way to maximize your
>signal-to-noise ratio is get the levels set correctly when you record
>the sound.  Otherwise, gain, normalization, or using the Gainer plugin
>will raise both signal and noise.

I don't think you get the gist of John's argument.  John explicitely 
says that using a Gainer plug-in is preferable to normalizing in the 
sample editor.  Saying that "they all raise signal and noise" misses 
the point, which is that anything done at 24 bits introduces more 
noise than when done at 32 bit.

Another but somewhat related argument against normalization: 
normalizing is imo as pointless as the digital zoom on digital 
cameras is.  If sufficient data is absent, the camera (Logic) can't 
invent it out of the blue.  With a 200 x 300 array, you can't make a 
_real_ 400 x 600 photo.  Likewise, if you have a recording that uses 
21 bits of dynamics, increasing that to 24 bits (normalizing by 
multiplying everything by 8) won't _really_ increase the resolution 
of your recording by a factor of 8.  If you then realise that it's 
not only pointless, but adds noise to boot, well... you get the 
picture.

-- 
Hendrik Jan Veenstra   h @ k n o w a r e . n l
Omega Art: http://www.omega-art.com/

Hendrik Jan Veenstra wrote:

>I don't think you get the gist of John's argument.  John explicitely 
>says that using a Gainer plug-in is preferable to normalizing in the 
>sample editor.  Saying that "they all raise signal and noise" misses 
>the point, which is that anything done at 24 bits introduces more 
>noise than when done at 32 bit.
>
I am actually referring to another point John made in his post....namely:

>Normalization doesn't affect the dynamic range of a signal, there will
>still be X dB difference between noisefloor and loudest peaks before
>and after normalization. 
>
That is, normalization raises both the noise floor (as well as the 
signal), but it also degrades the sound quality because it is a 
destructive edit made at a bit depth of 16 or 24-bit which is certainly 
introduces more quantization error than the 32-bit floating point 
resolution of the Gainer and Normalization function. 

Kind regards,
Dave

> ** Original Message follows... 

>    From: Hendrik Jan Veenstra <h@...>
> Subject: Re: Normalize Vs Gain
> 
> On a fine day, 12-03-2004, drummerboy5142 wrote:
> 
> >Like to end this argument once and for all. Is using the gain function
> >in the audio window better than normalizing your individual audio
> >tracks to bring the levels up to get the best quality recording to mix
> >with?
> 
> The best way (except for recording at decent levels :) is to use the 
> Gainer plug-in.  Since it's a plug-in, it operates in Logic's native 
> 32-bit space.  All destructive edits, such as normalizing, work on a 
> 24-bit level and introduce quantization errors (noise).
> 
> amgshaffer quoted John Pitcairn and wrote:
> >  >In Logic, it is far preferable to increase level by boosting the fader
> >>level and/or inserting Gainer plugins, which ensures that the increase
> >>in level is done both non-destructively and at a much more accurate
> >>32-bit floating point resolution. The result when added up across many
> >  > tracks is _clearly_ superior to normalizing all the audio.
> >
> >Given the above comment, I would assume the best way to maximize your
> >signal-to-noise ratio is get the levels set correctly when you record
> >the sound.  Otherwise, gain, normalization, or using the Gainer plugin
> >will raise both signal and noise.
> 
> I don't think you get the gist of John's argument.  John explicitely 
> says that using a Gainer plug-in is preferable to normalizing in the 
> sample editor.  Saying that "they all raise signal and noise" misses 
> the point, which is that anything done at 24 bits introduces more 
> noise than when done at 32 bit.
> 
> Another but somewhat related argument against normalization: 
> normalizing is imo as pointless as the digital zoom on digital 
> cameras is.  If sufficient data is absent, the camera (Logic) can't 
> invent it out of the blue.  With a 200 x 300 array, you can't make a 
> _real_ 400 x 600 photo.  Likewise, if you have a recording that uses 
> 21 bits of dynamics, increasing that to 24 bits (normalizing by 
> multiplying everything by 8) won't _really_ increase the resolution 
> of your recording by a factor of 8.  If you then realise that it's 
> not only pointless, but adds noise to boot, well... you get the 
> picture.
> 

Good topic. Despite knowing this, I sometimes use the gain function to bring up a word or phrase or reduce a sharp peak that's preventing me from raising my average level. I really wish the Gainer plug-in had a finer resolution - at least .5dB steps. Also, the automation is still very fiddly - sometimes the display will not show you your level and you have to look at the info bar and compare numbers with earlier automation points. BTW, why is this OT?

Mars

On a fine day, 13-03-2004, mars@... wrote:

>  >[...]
>Good topic. Despite knowing this, I sometimes use the gain function 
>to bring up a word or phrase or reduce a sharp peak that's 
>preventing me from raising my average level. I really wish the 
>Gainer plug-in had a finer resolution - at least .5dB steps.

Yes, couldn't agree more (about the resolution, that is).  Holds for 
the entire automation-stuff btw: why does the LC offer 10-bit 
resolution while hyperdraw and environment faders are still limited 
to this stupid stone-age 7-bit resolution?

>BTW, why is this OT?

Because we like a cosy small-scale chat amongst ourselves and would 
rather not bother the LUG-masses complaining about delayed delivery 
dates with sensible discussions :-).

-- 
Hendrik Jan Veenstra   h @ k n o w a r e . n l
Omega Art: http://www.omega-art.com/

> ** Original Message follows... 

>    From: Hendrik Jan Veenstra <h@...>
> Subject: re: Normalize Vs Gain
> 
> On a fine day, 13-03-2004, mars@... wrote:
> 
ffer 10-bit 
> resolution while hyperdraw and environment faders are still limited 
> to this stupid stone-age 7-bit resolution?
> 
> >BTW, why is this OT?
> 
> Because we like a cosy small-scale chat amongst ourselves and would 
> rather not bother the LUG-masses complaining about delayed delivery 
> dates with sensible discussions :-).
> 
:-) :-) Ah... it all makes sense now :-)

BTW, I can't remember if I ever responed to you, but both versions of Session Setup X worked fine - even under Panther.

Cheers

Mars

On a fine day, 14-03-2004, mars@... wrote:

>BTW, I can't remember if I ever responed to you, but both versions 
>of Session Setup X worked fine - even under Panther.

Ah, wonderful!  No, you never responded -- and I forgot about it as 
well :-).  So I suppose I can put it up for grabs then...  Thanks for 
beta-testing.  Still I think I should try myself one of these days 
(now that I've got an X-laptop).  I wondered (and forgot to ask you) 
if the 'labels' stuff simply carries over from 9 into X.3 or if that 
requries a new implementation (or omission -- whichever of the 2 is 
easiest :-).

-- 
Hendrik Jan Veenstra   h @ k n o w a r e . n l
Omega Art: http://www.omega-art.com/