Lord Hasenpfeffer wrote:

Thank God! I think you've finally and quite clearly defined the
problems we're having in this discussion!

wrote:

I don't know if you guys are having a semantic argument here, but from
my perspective, the very act of suppressing a peak IS clipping.

OK, this makes perfect sense.

Except that it is not the use of the word that is standard
across the recording industry. Clipping is what happens
when you boost the level without regard to what happens to
the peaks. They get "clipped" off flat at 0 dB full scale.

Normalization is boosting the level by the amount that will
bring the the highest peak in the region or the file to 0 dB
full scale.

Limiting is compression operation which changes the
dynamics. The level is not changed but instead the peaks
are held below a limit by a process of applying negative
gain by an amount that's a function of the momentary signal
level so as to keep the peaks below a given limit without
clipping them off. If it is too aggressive it will
introduce audible harmonic and intermodulation distortion.
Following that you can increase the level by the amount of
the limit to normalize. There should be a standard term for
increasing the gain while applying a 0 dB full scale limit
but there isn't that I've ever seen.


This "wrap-around" phenomenon is what I have so far been referring to as
"clipping".

I've seen "wrapping" used. You don't see that much any
more. I'm surprised Audacity retains it. It can speed the
performance of increasing the level though by eliminating
the need for an overflow check in the multiplication loop
but few DAW's still have that option.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein

Bob Cain wrote:

Except that it is not the use of the word that is standard across the
recording industry. Clipping is what happens when you boost the
level without regard to what happens to the peaks. They get
"clipped" off flat at 0 dB full scale.

Yes, thank you. As of earlier today I have now correctly adopted this
term to describe that phenomenon.

Normalization is boosting the level by the amount that will bring the
the highest peak in the region or the file to 0 dB full scale.

OK, I've been using the term "normalization" to describe *everything* I
do with the "normalize" application. I was not aware that the term
"normalization" applies *only* to when you bring the highest peak to
0dB. That's what the application does to a WAV *by default* with no
additional, custom level or gain adjustment specifications.

There should be a standard term for increasing the gain while
applying a 0 dB full scale limit but there isn't that I've ever seen.

How about "over-normalization" (for lack of anything better)?

So when I opt to "over-normalize" the level of a WAV to my preferred
"2dB hotter" setting than what the application would do by default, I am
no longer "normalizing" the WAV but doing something else to it instead
which really doesn't have a name. Hmmm...

Limiting is compression operation which changes the dynamics. The
level is not changed but instead the peaks are held below a limit by
a process of applying negative gain by an amount that's a function of
the momentary signal level so as to keep the peaks below a given
limit without clipping them off. If it is too aggressive it will
introduce audible harmonic and intermodulation distortion. Following
that you can increase the level by the amount of the limit to
normalize.

So if, by definition, "limiting" requires no change in level - and only
offers peak suppression at some artificial threshold, wouldn't this
"over-normalization" thing that I've been doing just be called
"compression" since it is a combination of "boosting the level" while
the "clip all peaks at 0dB" rule?

I've seen "wrapping" used.

OK, so is "wrapping" the officially factory-authorized industry standard
term that's used to describe that effect?

I had previously been calling *that* effect "clipping" because when you
hear it it kinda makes a harsh, audible "clipping noise"! I thought
"clipping" was in reference to the sound that "wrapping" makes, not to
the process of flattening the peaks at 0dB.

You don't see that much any more.

It doesn't seem to have much use even as a rude sound effect.

I'm surprised Audacity retains it.

My wife asked me today what possible purpose it could serve - and I
could not provide an answer.

It can speed the performance of increasing the level though by
eliminating the need for an overflow check in the multiplication loop
but few DAW's still have that option.

Yeah, maybe that could have been useful in the stone age but not any longer.

It could* be useful if you just want to do a quick test to see roughly
how many peaks are gonna get clipped if you do a certain experimental
thing with your WAV. Then if you don't like that result you could
"undo" and try again repeatedly until you get it way you want it.

Maybe?

Myke

Bob Cain wrote:

How about "over-normalization" (for lack of anything better)?

Nah, too long. :-)

Yah, that's what I figgered you were gonna say.

Well, it can be thought of as limit-normalize so how about
"limitize"? :-)

Aha! I like that. Limitize it is!! You readin' this, Geoff???? We're
gonna start callin' it "limitize"! And you don't even have to read a
book to understand what we're sayin'!

"Compression" is a gentler form of limiting. The process
is actually nearly the same. Limiting is compression with
a certain form of gain function.

Hmmm... The way I've heard it described in the past, I'd think
compression was more "violent" than limiting because, as I understand
it, it involves both "lowering the ceiling" *and* "raising the floor"
whereas limiting only affects things topside.

[Wrapping is] so rare now that nobody calls it much of anything.

Ha! Rare for you maybe. I've encountered it a *lot* because I've
always been working with the "Don't allow clipping" box checked!

I use "undo" for that kind of thing all the time. I'm not
sure the speed gained by not checking for overflow is worth
the effort of turning it on and off.

No, I wouldn't do it just for the speed increase. I just think that
during test runs prior to undo's, the wrapped values would make it
easier to visually locate ranges containing peaks that would otherwise
be subject to clipping in the other mode. I mean, "wrapped" samples
make for some pretty big "mountains" and "valleys" in the middle of some
otherwise mere "rocky terrain".

Myke

Pete Carney wrote:

Myke, please go study what clipping, normalization, compression, limiting
and expansion are.

Please go view this screenshot!

http://www.mykec.com/mykec/images/20...ey_Smoking.png

Myke

--

-================================-
Windows...It's rebootylicious!!!
-================================-

Who after seeing this screenshot will be the first to report back and
honestly tell me that my digitally remastered version of "Dark Side Of
The Moon" is not sonically superior to Mobile Fidelity Sound Lab's own
original, gold-plated offering???

http://www.mykec.com/mykec/images/20...ey_Smoking.png

If this is the kind of shoddy product that MFSL continally puts out to
consumers, I know *I'll* be never buying anything with their name on it
again. Pink Floyd, "Dark Side Of The Moon" is hardly a trivial product!

Myke

--

-================================-
Windows...It's rebootylicious!!!
-================================-

Lord Hasenpfeffer wrote:

Who after seeing this screenshot will be the first to report back and
honestly tell me that my digitally remastered version of "Dark Side Of
The Moon" is not sonically superior to Mobile Fidelity Sound Lab's own
original, gold-plated offering???


http://www.mykec.com/mykec/images/20...ey_Smoking.png

If this is the kind of shoddy product that MFSL continally puts out to
consumers, I know I'll be never buying anything with their name on it
again. Pink Floyd, "Dark Side Of The Moon" is hardly a trivial product!


Your screenshot is useless. What did the normalize tell about the peak
value of the disk, and what was the adjustment? And what about audacity
screenshot about those two milliseconds where the peak is?

Teemu

Teemu Torma wrote:

Your screenshot is useless.

The purpose of that screenshot is to demonstrate how - over the course
of an entire album - I have sweetly extended and not at all reduced the
dynamic range of the original audio WAV. To that end, my screenshot is
not useless. As for your other questions about it, however, I can see
your point.

What did the normalize tell about the peak value of the disk,

I'm sure you'll have a really hard time believing this but, I forgot to
re-analyze the resulting WAV for this information before I deleted it.

But I'd be perfectly happy to exactly repeat the process and provide you
with the answer if you'd like.

and what was the adjustment?

+4.5dB.

And what about audacity screenshot about those two milliseconds where
the peak is?

Behold:

http://www.mykec.com/mykec/images/20...FSL_Zoom_1.png
http://www.mykec.com/mykec/images/20...FSL_Zoom_2.png

These are exploded views of the peak which occurs at approximately the
33 minute mark in my remastered version.

As you can clearly see, I have only improved upon the original WAV form.
In this test, I can detect no damage whatsoever to the integrity of
the original information; only dramatic improvment.

BTW, after normalizing MFSL's original WAV, I compared my results with
that of Capitol's 1994 digitally remastered reissue and found remarkable
similarities.

And that is why I routinely use "normalize" to -10dBFS.

Myke

--

-================================-
Windows...It's rebootylicious!!!
-================================-

"Lord Hasenpfeffer" wrote in message
...
Who after seeing this screenshot will be the first to report back and
honestly tell me that my digitally remastered version of "Dark Side Of
The Moon" is not sonically superior to Mobile Fidelity Sound Lab's own
original, gold-plated offering???


http://www.mykec.com/mykec/images/20...ey_Smoking.png

If this is the kind of shoddy product that MFSL continally puts out to
consumers, I know *I'll* be never buying anything with their name on it
again. Pink Floyd, "Dark Side Of The Moon" is hardly a trivial product!


I find it hard to believe that you, especially after all the help offered,
still totally miscomprehend almost everything about music and normalisation.

Do you really think that every *track* is somehow deficient if it isn't the
same loudness as every other track on a CD, and that by normalising it
(whatever method) that you are somehow being clever, or improving the music
(or the dymnamic range).

In fact I can't believe that anybody could be that dumb, and that you must
actually be a USENET troll.

Bye

geoff

Geoff Wood wrote:

Do you really think that every track is somehow deficient if it isn't the
same loudness as every other track on a CD, and that by normalising it
(whatever method) that you are somehow being clever, or improving the
music (or the dymnamic range).

In fact I can't believe that anybody could be that dumb, and that you must
actually be a USENET troll.

For Myke's defend I must say that he seems to normalize the whole disk at
once, not individual songs.

Teemu

Geoff Wood wrote:

I find it hard to believe that you, especially after all the help offered,
still totally miscomprehend almost everything about music and normalisation.

I understand that I am using a program called "normalize" that obviously
does more than just normalize. That's why I tend to enclose the word
"normalize" in quotation marks so often - because I'm aware that it's
not really just normalizing my WAVs.

Do you really think that every *track* is somehow deficient

No. Most every modern (i.e. 1994-present) CD and/or 24-bit digitally
remastered CD in my library requires no "munging" (as you say) - because
their peaks and levels are already the same or "better" than what I
would be able to make them by "normalize"-ing them to -10dBFS. Most
every other CD that I own, however, does need help, yes.

Andrew Lloyd Webber's 1987 "Phantom Of The Opera" CD set is a
particularly notable exception. When I scanned its peaks and levels
with "normalize" I was pleasantly surprised to find that it was already
"perfect". Not bad, I thought, for a CD I've owned since the late 80s
when it was practically still new.

if it isn't the same loudness as every other track on a CD,

Look again at my screenshot. There are *many* alternating loud and soft
passages in "Dark Side". They are of "the same loudness" at all. Where
the music should be subtle it remains subtle. Where it should be
dramatic, it is still dramatic.

It's the MFSL original WAV that is more consistently loud from start to
finish. Their WAV is damn near arrow straight!

and that by normalising it (whatever method) that you are somehow
being clever,

Nah. Not even close.

or improving the music (or the dymnamic range).

Improving the music? Nah. Nobody beats the Floyd.

Improving the dynamic range? I believe so, yes. Is that a bad thing?
Yesterday everyone was telling me how wrong it is to reduce dynamic
range - to which I easily agree. I've never heard anybody complain
about having too much dynamic range.

Back when CDs were new I used to read all the great things about how
it's dynamic range is something like +/-90dB or thereabouts, while that
of the lowly vinyl LP was something like +/-27dB. And there was dancing
in the streets.

In fact I can't believe that anybody could be that dumb, and that you must
actually be a USENET troll.

Please do not reduce yourself to name-calling. I am not a troll.

Myke

--

-================================-
Windows...It's rebootylicious!!!
-================================-

"Teemu Torma" wrote in message
...

In fact I can't believe that anybody could be that dumb, and that you
must
actually be a USENET troll.

For Myke's defend I must say that he seems to normalize the whole disk at
once, not individual songs.


If you look at some of his posts, that is exactly what he is doing. However
he seems to have moderated this slightly wrt his latest DSOTM slant, though
he is still thinking that 4dB difference in peak level is somehow the
difference between caviar and ****.

He also refuses to comprehend that such recordings' dynamic range is limited
by the original source tapes, and that it is not necessary (or desirable)
for every piece of music to utilise all the 96dB dynamic range available on
CD. He has had it explained, been pointed in the direction of clear concise
factual explanations, and still refuses to attempt to comprehend where he is
totally missing the point.

Such stubborness is beyond the realms of likelihood, especially considering
the effort he (misguidedly) puts into his music , and I suspect he is merely
posting these ridiculous outrageous things to prompt responses and get peope
het up (trolling). Then he tries to exacerbate things by starting an
OS-religous war.

Naaa, I give up.

geoff

Teemu Torma wrote:

For Myke's defend I must say that he seems to normalize the whole disk at
once, not individual songs.

THANK YOU, TEEMU! You've got it!

When I rip 10 songs from 1 original CD, I "batch normalize" the entire
set to preserve their original, relative loudnesses in relation to each
other. This I have already explained in this forum on more than four
occasions. I thought this was also highly apparent in the screenshots I
have provided - which is exactly the reason why my initial screenshot
displays the WAV of the *entire* album and not just a single song from
it. Geez.

Onion other hand, whenever I seek to create a "mix-CD" compilation of
songs from various, unrelated sources, I individually "normalize" each
track to create a more-or-less consistent loudness across the entire mix.

Some, however, still insist on ignoring these facts and in so doing feel
justified to label me a "Liniot" and a "****wit" and even a "USENET troll".

Myke

--

-================================-
Windows...It's rebootylicious!!!
-================================-