[Roy W. Rising[_2_]]

(Scott Dorsey) wrote:
William Sommerwerck wrote:
"geoff" wrote in message
...
On 25/12/2014 9:28 a.m., William Sommerwerck wrote

When I made live recordings, I'd ask the orchestra to play the
loudest passage from the works they were performing, then set the
level about 1dB below that. My experience was that orchestras play a
bit louder during the actual performance.

That's an incredibly narrow margin to set! If they play what you
notice to be "a bit louder", then they are well over the 1dB.

I'm talking about what the meters showed, not what I heard.


My inclination is to set it to "the loudest passage" is at -18dBFS.
Because when they actually play it for real, it will be 6dB louder than
they said it would be in performance, and then add 12dB worth of safety
margin for something else louder that wasn't forseen.
--scott

Back in the days of steam-powered videotape, ABC-TV Labs did some extensive
research on this topic. It was found that, for a wide range of program
material, the highest peak factor was about 16dB above Zero VU. Systems
design recommendations were that the least headroom should be 18dB. To
that, sometimes, was added something called a "crest factor" of 6dB.
Program Amps capable of +24dBm output made this feasible in some
circumstances. Then came Program Amps capable of +30dBm output (1 watt!).
Systems became cleaner and operators grew lazier! Oh well!

The first generation of Pro Tools gave us Zero = only -14dBFS. At the
same time, the -20dBFS was becoming standardized in most digital systems.
Instead of letting us "turn down" 6dB ahead of the recorders, a mindless
manager imposed limiters, even though operators were able to identify and
complain about what it did to their mixes. Again, Oh well!

--
~ Roy
"If you notice the sound, it's wrong!"

[William Sommerwerck]

"Roy W. Rising" wrote in message ...

Back in the days of steam-powered videotape, ABC-TV Labs did
extensive research on this topic. It was found that, for a wide
range of program material, the highest peak factor was about 16dB
above Zero VU. Systems design recommendations were that the
least headroom should be 18dB. To that, sometimes, was added
something called a "crest factor" of 6dB.

Some instruments -- such as piano and trumpet -- have crest factors
(instantaneous peak to average) of 16dB. This is for a solo instrument.
Ensembles would tend to have lower crest factors.

[Roy W. Rising[_2_]]

"William Sommerwerck" wrote:
"Roy W. Rising" wrote in message
...

Back in the days of steam-powered videotape, ABC-TV Labs did
extensive research on this topic. It was found that, for a wide
range of program material, the highest peak factor was about 16dB
above Zero VU. Systems design recommendations were that the
least headroom should be 18dB. To that, sometimes, was added
something called a "crest factor" of 6dB.

Some instruments -- such as piano and trumpet -- have crest factors
(instantaneous peak to average) of 16dB. This is for a solo instrument.
Ensembles would tend to have lower crest factors.

I suspect the parlance was incorrect. The 16dB peak factor included what
you describe. Perhaps the extra headroom was in defense of operators who
could not keep a VU meter out of the red, where there is some forgiveness.
Later came Peak Program Meters. The same operators could not keep them out
of the Yellow, where there is much less forgiveness! I classified the two
types as Meter Minders and Pin Pounders.

--
~ Roy
"If you notice the sound, it's wrong!"

[Bob Quintal]

wrote in
:

Trevor wrote: "It's usually in post the decisions are made to
compress and clip the hell out of it to maximise the apparent
loudness and match what all the other idiots are doing"


At least there doing so ensures all the bits are being used, on a
16 or 24bit level (!)

Only a psychotic would worry about not using every bit of a recorder's
range.



--
Bob Q.
PA is y I've altered my address.

[Peter Larsen[_3_]]

"William Sommerwerck" skrev i en meddelelse
...

"Roy W. Rising" wrote in message
...

Back in the days of steam-powered videotape, ABC-TV Labs did
extensive research on this topic. It was found that, for a wide
range of program material, the highest peak factor was about 16dB
above Zero VU. Systems design recommendations were that the
least headroom should be 18dB. To that, sometimes, was added
something called a "crest factor" of 6dB.

Some instruments -- such as piano and trumpet -- have crest factors
(instantaneous peak to average) of 16dB. This is for a solo instrument.

Not correct, I posted a lot about this some years ago, and some care is
required to get the observed crestfactor to actually be the crest factor of
the instrument observed and not of the recording of the instrument. A
reasonable asumption is that a sane recording of acoustic instruments ends
up around a crest factor of 25 dB.

A recording of a string quartet in a reverberant hall - yes, the hall from
the video of the piano trio I posted a link to some time ago - will have a
crest factor around 27 dB.

Generally it is time to ask questions in case you observe crest factors
below 25 dB in a recording. It may be bass heavy, it may be made with long
microphone to source distance in a reverberant room or it may be clipped or
compressed.

If we are talking single tracks of a classical music recording made with mic
to sound source distance in the 1 to 5 feet range the crest factor to allow
for is 35 dB.

Ensembles would tend to have lower crest factors.

Your data differ drastically from my observations.

Kind regards

Peter Larsen

[Peter Larsen[_3_]]

reposting because my post via astraweb vanished ...

"William Sommerwerck" skrev i en meddelelse
...

"Roy W. Rising" wrote in message
...

Back in the days of steam-powered videotape, ABC-TV Labs did
extensive research on this topic. It was found that, for a wide
range of program material, the highest peak factor was about 16dB
above Zero VU. Systems design recommendations were that the
least headroom should be 18dB. To that, sometimes, was added
something called a "crest factor" of 6dB.

Some instruments -- such as piano and trumpet -- have crest factors
(instantaneous peak to average) of 16dB. This is for a solo instrument.

Not correct, I posted a lot about this some years ago, and some care is
required to get the observed crestfactor to actually be the crest factor of
the instrument observed and not of the recording of the instrument. A
reasonable asumption is that a sane recording of acoustic instruments ends
up around a crest factor of 25 dB.

A recording of a string quartet in a reverberant hall - yes, the hall from
the video of the piano trio I posted a link to some time ago - will have a
crest factor around 27 dB.

Generally it is time to ask questions in case you observe crest factors
below 25 dB in a recording. It may be bass heavy, it may be made with long
microphone to source distance in a reverberant room or it may be clipped or
compressed.

If we are talking single tracks of a classical music recording made with mic
to sound source distance in the 1 to 5 feet range the crest factor to allow
for is 35 dB.

Ensembles would tend to have lower crest factors.

Your data differ drastically from my observations.

Kind regards

Peter Larsen

[John Williamson]

On 25/12/2014 17:29, Bob Quintal wrote:
wrote in
:

Trevor wrote: "It's usually in post the decisions are made to
compress and clip the hell out of it to maximise the apparent
loudness and match what all the other idiots are doing"


At least there doing so ensures all the bits are being used, on a
16 or 24bit level (!)

Only a psychotic would worry about not using every bit of a recorder's
range.



In the early days of digital, when some converters only had an effective
resolution of 14 bits, and poor linearity at low levels, it was more
important than it is now.

--
Tciao for Now!

John.

[Les Cargill[_4_]]

Mike Rivers wrote:
On 12/25/2014 3:12 AM, geoff wrote:
You do realise that 1dB is barely perceivable change in level. So they
really didn't play any louder. I would expect a likely increase of
maybe 6dB in the the heat of a typical real performance.

I always believed that, but then I started reading about mastering
engineers adding a couple of tenths of a dB here and there and made the
client smile.



That might be in the domain of underdressed emperors. Then again, it
takes a heck of a customer to appreciate doing nothing.

--
Les Cargill

[Peter Larsen[_3_]]

You can add anything in post but clarity - that you can only loose
and sound recording and the entire post-processing until release
is about loosing as little as possible.

I think you mean "lose" and "losing", unless things are just kinda
rattling around. (Caught those because I make errors like that.)

Auw. I have erred. After 30 years of usenetting I have posted an actual and
misleading spelling error. About time ... O;-) ... you are quite right,
loose is what happens to the stereo image if the pair is just a wee bit too
distant from the ensemble for the mic setup chosen. But was that a reason
for breaking your omerta?!?`- about the only secret word you haven't used is
the one that start with a B as in Beethoven and they even make an object
called a composer.

Actually, in the clarity department, minor miracles can often
be had by thoughtful and careful cuts of varying degrees in the
lower octaves - LF and Lo MF dips, varying shapes of high pass,
etc. This could be applied to troublesome tracks or to the
entire mix.

And these issues are the ones where FFT analysis and a hand-drawn FFT EQ
curve really can do wonders, also because those lower frequency range
oddities often can be assumed to NOT be minimum phase, so the use of a
"phase linear" FFT equalizer to address them is a very context relevant
choice.

Surprisingly, you can often "infer" clarity by adding HF energy
-- not in the form of EQ, but rather with HF-heavy
reverb done to taste (and with the appropriate pre-delay).

Not to worry, hardly anybody has grasped that more pre-delay and less verb
most of the time will be better than less pre-delay and a lot more - AND
THEN SOME - reverb.

Finally, a few processing tools can also aid, such as SPL's
Vitalizer or Transient Designer. Neither can create
something from nothing, and both are easily overused, but
they can help remove at least one layer of murk.

Like prefab in a kitchen, apply with care, start with doing half of half of
what you think is needed, and tell no-one.

Frank
Mobile Audio

Oh, you asked for examples of my recordings, I'll post links if/when some
people make them available as they said they would ...

Kind regards

Peter Larsen

[Mike Rivers[_2_]]

On 12/25/2014 7:25 PM, Scott Dorsey wrote:
I don't consider level setting to be a precise thing that you can (or
should) do with a measurement. It's a thing where a measurement can
guide you in using your good judgment.

This was surely the case in the analogue world, but I disagree that it remains
the case in the digital world where the tonal quality does not change at all
until you hit clipping, and then it changes a lot.

Where the good judgment comes in is in deciding, once you see the level
that the meters are showing for peaks, whether to leave it there, boost
it a little because you know darn well that it won't get louder than
that, or back it off because you don't know the music and you might be
dozing during a crescendo and not be aware that it's coming.

--
For a good time, visit http://mikeriversaudio.wordpress.com

[William Sommerwerck]

Some instruments -- such as piano and trumpet -- have crest factors
(instantaneous peak to average) of 16dB. This is for a solo instrument.

I don't know whether I read this or measured it (on a 'scope).

I'm willing to believe that some instruments have a crest factor of 25dB. But
a flute sure doesn't. Whereas piani and trumpets have spiky waveforms.

Can you point us to a reference?

[Nil[_2_]]

On 25 Dec 2014, wrote in rec.audio.pro:

Trevor wrote: "It's usually in post the decisions are made to
compress and clip the hell out of it to maximise the apparent
loudness and match what all the other idiots are doing"


At least there doing so ensures all the bits are being used, on a
16 or 24bit level (!)

If you ever learn to quote in a legible manner you might earn some more
respect.

[Peter Larsen[_3_]]

"William Sommerwerck" skrev i en meddelelse
...

Some instruments -- such as piano and trumpet -- have crest factors
(instantaneous peak to average) of 16dB. This is for a solo instrument.

I don't know whether I read this or measured it (on a 'scope).

Translates to "I think it is so, and thus claim it as constituting A fact".

I'm willing to believe that some instruments have a crest factor of 25dB.
But a flute sure doesn't. Whereas piani and trumpets have spiky waveforms.

Can you point us to a reference?

Bing found it instantly with the search term "crest factor peter larsen
rec.audio.pro":

https://groups.google.com/forum/#!to...ch/LlVr_MmQkeg


Kind regards

Peter Larsen

[Peter Larsen[_3_]]

"Peter Larsen" skrev i en meddelelse
k...

"William Sommerwerck" skrev i en meddelelse
...

Some instruments -- such as piano and trumpet -- have crest factors
(instantaneous peak to average) of 16dB. This is for a solo instrument.

I don't know whether I read this or measured it (on a 'scope).

Translates to "I think it is so, and thus claim it as constituting A
fact".

I'm willing to believe that some instruments have a crest factor of 25dB.
But a flute sure doesn't. Whereas piani and trumpets have spiky
waveforms.

Can you point us to a reference?

Bing found it instantly with the search term "crest factor peter larsen
rec.audio.pro":

https://groups.google.com/forum/#!to...ch/LlVr_MmQkeg

Oh, and this one too, the data-set is probably three dB off because of a
cool edit odditity explained he

https://groups.google.com/forum/#!to...nd/q579MxeVrqg

Kind regards

Peter Larsen

[hank alrich]

Bob Quintal wrote:

wrote in
:

Trevor wrote: "It's usually in post the decisions are made to
compress and clip the hell out of it to maximise the apparent
loudness and match what all the other idiots are doing"


At least there doing so ensures all the bits are being used, on a
16 or 24bit level (!)

Only a psychotic would worry about not using every bit of a recorder's
range.

Maybe some get paid by the bit. ;-)

--
shut up and play your guitar * HankAlrich.Com
HankandShaidriMusic.Com
YouTube.Com/WalkinayMusic

[geoff]

On 26/12/2014 2:55 a.m., Mike Rivers wrote:
On 12/25/2014 3:07 AM, geoff wrote:
A peak reading meter is the ONLY way you can accurately set levels.

What does "accurately set levels" mean to you? How accurate do you want
to be? Do you want to be sure that at some place the level reaches full
scale? Or do you want to be sure that at no place does it ever reach
full scale, at least for content that you want to keep?


One needs to know that the peak level does not get 0dBFS. Which you
won't achieve with a ballistic-weighted meter.

And if one's policy may be to set peaks at (say) -12dB, then you can set
that with confidence.

geoff

[geoff]

On 26/12/2014 4:14 a.m., wrote:
Trevor wrote: "It's usually in post the decisions are made to compress and clip the hell out of it to maximise the apparent loudness and match what all the other idiots are doing"


At least there doing so ensures all the bits are being used, on a 16 or 24bit level (!)



In 24 bit it's just not an issue.

Tracking at 16 bit is pretty uncommon for the last 10 years(ish), unless
you are recording on a DAT, DA-88, or older ADAT.

geoff

[geoff]

On 26/12/2014 3:00 a.m., Mike Rivers wrote:
On 12/25/2014 3:12 AM, geoff wrote:
You do realise that 1dB is barely perceivable change in level. So they
really didn't play any louder. I would expect a likely increase of
maybe 6dB in the the heat of a typical real performance.

I always believed that, but then I started reading about mastering
engineers adding a couple of tenths of a dB here and there and made the
client smile.



...... and the ones who did nothing at all, said they did, and made the
clients smile.


geoff

[John Williamson]

On 26/12/2014 09:04, geoff wrote:
On 26/12/2014 3:00 a.m., Mike Rivers wrote:
On 12/25/2014 3:12 AM, geoff wrote:
You do realise that 1dB is barely perceivable change in level. So they
really didn't play any louder. I would expect a likely increase of
maybe 6dB in the the heat of a typical real performance.

I always believed that, but then I started reading about mastering
engineers adding a couple of tenths of a dB here and there and made the
client smile.



..... and the ones who did nothing at all, said they did, and made the
clients smile.


Who was it that had a dummy knob on the desk for the client to adjust stuff?

Apparently, it caused smiles to break out all round when it was turned.

--
Tciao for Now!

John.

[Mike Rivers[_2_]]

On 12/26/2014 10:02 AM, geoff wrote:
In 24 bit it's just not an issue.

It could be, depending on where you can turn the level down to stay
within your margin of choice. If you have a very clean signal going into
a 24-bit A/D converter, you can bring peaks up to full scale without any
significant damage. But if the only knob you have available to reduce
the record level is the input gain of you could be compromising the S/N
ratio right at the front. And when you add 10 dB to the recording to get
the peaks where the client wants them, you'll be amplifying the preamp
noise.

This may only be a theoretical problem, but it's not good to ignore any
part of the system when choosing a record level.

--
For a good time, visit http://mikeriversaudio.wordpress.com

[Peter Larsen[_3_]]

"John Williamson" skrev i en meddelelse
...

Who was it that had a dummy knob on the desk for the client
to adjust stuff?

Apparently, it caused smiles to break out all round when it
was turned.

Ah, you have never adjusted a reverb in bypass mode until it was just right?

John

Kind regards

Peter Larsen

[Arny Krueger[_5_]]

One word: eBay.

"mcp6453" wrote in message
...

With all due respect to Coleman Audio, can someone suggest a less expensive
way to find a pair of analog VU meters for a
stereo program buss?

http://goo.gl/6mWuGZ

I don't even see how big the meters are. Maybe I have to roll my own.

[Scott Dorsey]

geoff wrote:
On 26/12/2014 2:55 a.m., Mike Rivers wrote:
On 12/25/2014 3:07 AM, geoff wrote:
A peak reading meter is the ONLY way you can accurately set levels.

What does "accurately set levels" mean to you? How accurate do you want
to be? Do you want to be sure that at some place the level reaches full
scale? Or do you want to be sure that at no place does it ever reach
full scale, at least for content that you want to keep?

One needs to know that the peak level does not get 0dBFS. Which you
won't achieve with a ballistic-weighted meter.

It can be! In the early digital days when we did not have budget for
full and accurate digital metering, I used an analogue meter with a peak
reading circuit driving it; the ballistics were such that it took some time
to reach the correct reading but the peak hold circuit driving it kept that
reading going to the meter for much longer.

I would, of course, not recommend such metering today. If you look at some
late 1980s postings from this newsgroup you can probably see Gabe Weiner
arguing that my arrangement was insufficiently accurate. However, all the
recordings I have from back then have the levels spot on.

And if one's policy may be to set peaks at (say) -12dB, then you can set
that with confidence.

What is your meter measuring? If it's designed to measure peaks, that is
one thing. If it is designed to measure something else, that is great, but
if so it's not of much use to measure peaks.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Scott Dorsey]

geoff wrote:
Tracking at 16 bit is pretty uncommon for the last 10 years(ish), unless
you are recording on a DAT, DA-88, or older ADAT.

I'm still frequently tracking to 16 bit, but I am actually tracking to real
16 bits with all the bits being good and valid. And the room noise floor is
still more of an issue than the converter noise or linearity.

Contrast that with the SV3700 of the eighties where if you were getting any
more than 12 valid signal-correlated monotonic bits I will eat my hat.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Scott Dorsey]

Peter Larsen wrote:

With my SV3800 sound would be cleanest if the upper bit was not left unused
and the inaudibility of short clipping relied on, if there actually was some
I unclipped the recording.

Which is horrible, but it was the nature of the SV3800. Those machines were
a good example of every possible way to screw up a converter design. They
sounded appreciably better at 48 ksamp/sec rather than 44.1, which led a lot
of people to surmise about audibility of utrasonics. In fact, the issues
were that the same anti-aliasing filter constants were used at both rates.

That's on top of the terrible converter linearity and the nonstandard I/O.
Panasonic used to send people to the AES standards committee meeting and
try to convince people to change the spec rather than fix their interfaces.

With that gamble it gets required to allow for musical genre. Very good
singing wimmen can be trusted when recording, I know of ONE (1) who sings
with so much precision and control that a compressor need not be deployed in
post, the rest of them can be trusted to always have an extra 10 dB of
loudness they come up with with or without relevance in the musical context.

Even then I ran out of downward bits at an avant-garde event. Usually it is
just a pair of eq and stereo tricks to lift musicians commentary and make it
audible in post, but with that recording I couldn't lift those passages as
much as had been relevant, the voice got robotic in character. Not totally
out of place in the musical context, even if purely by natural instruments,
but there also was no real room ambience left.

With the SV3700 this happens only 40dB down. Mind you, if you'd been running
a 1/4" 2-track machine the noise would have been building up at that point
too.

You could have put a compressor in-line and makeup gain to
keep the level up to within one bit of full-scale, that is,
unless this recording was 'in the field'.

Do you read what people post? - William refers to making live recordings.
Also William is into hifi and possibly knows just how much mess a compressor
makes with the sound and what the cost is in clarity.

Actually, a safety limiter can be a real help in this kind of situation. You
don't hit the limiter very often... you don't hit it ever if you can avoid it.
But it's better to hit the limiter than hit the digital brickwall. Pretty
much all the portable recorders intended for film work had some sort of
mildly inoffensive safety limiter that could be enabled if needed.

You can not add clarity, you have to have it and preserve it. Which is why
risking a couple of dB of clipping on the (modded) SV3800 worked to the
recordings advantage. Sadly replacing opamps and capacitor in it was only
about 1/3 the required rebuild O:-) - it actually became OK as long as the
uppermost bit was used, but moving to 24 bit recording was an ambience
revelation.

To be honest you could just have thrown a Symetrix 620 in front of that 3800
and been surprisingly happy. Moving to real 16 bit recording would have been
an ambience revelation!
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[mcp6453[_2_]]

On 12/22/2014 5:01 PM, PStamler wrote:
On Monday, December 22, 2014 2:54:54 PM UTC-6, Scott Dorsey wrote:
mcp6453 wrote:
With all due respect to Coleman Audio, can someone suggest a less expensive way to find a pair of analog VU meters for a
stereo program buss?

http://goo.gl/6mWuGZ

I don't even see how big the meters are. Maybe I have to roll my own.

There really aren't many. The Dorrough is even more money. You can
find surplus meters and add resistors and metalwork yourself.

I think Simpson will still sell you a real VU meter movement for $250.
--scott

Which means you're probably better off cannibalizing them from someplace.

Two other possibilities: National Semiconductor, in their datasheets for the LED-array drivers which power a lot of peak-reaing meters, gave a circuit which supposedly emulated BU meter ballistics. I have no idea how well it worked.

And Heathkit claimed that the meter on their Audio Analyzer (an SMPTE IM meter) had true-VU ballistics. I've found it useful as a good stand-in. Of course, for stereo you'd need two of them -- plus a way to line up your 0 VU point to the boxes'.

As much as I hate LED VU meters, this one looks great. Unfortunately, the guy who made it is MIA from YouTube.

http://www.youtube.com/watch?v=5P1fNMrfxMo

This one looks nice, but these guys are in Canada, and they insist on charging $25 to ship this tiny kit to the US. Even
then, you have to find a way to mount it in a case or behind a panel.

http://www.augustica.com/dejavu-c-17...ter-dejavu-p-8

It looks like the Augustica guys may have gotten the design from this guy in CZ:

http://s-o.webnode.cz/vu-metr/

I may try to order some from him, if I can figure out how.

[John Williamson]

On 26/12/2014 18:31, mcp6453 wrote:
On 12/22/2014 5:01 PM, PStamler wrote:
And Heathkit claimed that the meter on their Audio Analyzer (an SMPTE IM meter) had true-VU ballistics. I've found it useful as a good stand-in. Of course, for stereo you'd need two of them -- plus a way to line up your 0 VU point to the boxes'.

As much as I hate LED VU meters, this one looks great. Unfortunately, the guy who made it is MIA from YouTube.

http://www.youtube.com/watch?v=5P1fNMrfxMo

This one looks nice, but these guys are in Canada, and they insist on charging $25 to ship this tiny kit to the US. Even
then, you have to find a way to mount it in a case or behind a panel.

http://www.augustica.com/dejavu-c-17...ter-dejavu-p-8

It looks like the Augustica guys may have gotten the design from this guy in CZ:

http://s-o.webnode.cz/vu-metr/

I may try to order some from him, if I can figure out how.

You don't need to. There's a circuit diagram and a link to the PIC
program on the page.

There's even a PCB layout.

--
Tciao for Now!

John.

[Les Cargill[_4_]]

Bob Quintal wrote:
wrote in
:

Trevor wrote: "It's usually in post the decisions are made to
compress and clip the hell out of it to maximise the apparent
loudness and match what all the other idiots are doing"


At least there doing so ensures all the bits are being used, on a
16 or 24bit level (!)

Only a psychotic would worry about not using every bit of a recorder's
range.




Maybe they run their car at redline all the time.

--
Les Cargill

[Les Cargill[_4_]]

geoff wrote:
On 26/12/2014 4:14 a.m., wrote:
Trevor wrote: "It's usually in post the decisions are made to compress
and clip the hell out of it to maximise the apparent loudness and
match what all the other idiots are doing"


At least there doing so ensures all the bits are being used, on a 16
or 24bit level (!)



In 24 bit it's just not an issue.

Tracking at 16 bit is pretty uncommon for the last 10 years(ish), unless
you are recording on a DAT, DA-88, or older ADAT.

geoff

It was possible to buy 16 bit gear 10 years ago, and a lot of that still
works.


--
Les Cargill

[geoff]

On 26/12/2014 11:51 p.m., Mike Rivers wrote:
On 12/26/2014 10:02 AM, geoff wrote:
In 24 bit it's just not an issue.

It could be, depending on where you can turn the level down to stay
within your margin of choice. If you have a very clean signal going into
a 24-bit A/D converter, you can bring peaks up to full scale without any
significant damage. But if the only knob you have available to reduce
the record level is the input gain of you could be compromising the S/N
ratio right at the front. And when you add 10 dB to the recording to get
the peaks where the client wants them, you'll be amplifying the preamp
noise.

This may only be a theoretical problem, but it's not good to ignore any
part of the system when choosing a record level.



..... assuming a 'sensible' margin here.

geoff

[hank alrich]

Mike Rivers wrote:

On 12/26/2014 10:02 AM, geoff wrote:
In 24 bit it's just not an issue.

It could be, depending on where you can turn the level down to stay
within your margin of choice. If you have a very clean signal going into
a 24-bit A/D converter, you can bring peaks up to full scale without any
significant damage. But if the only knob you have available to reduce
the record level is the input gain of you could be compromising the S/N
ratio right at the front. And when you add 10 dB to the recording to get
the peaks where the client wants them, you'll be amplifying the preamp
noise.

This may only be a theoretical problem, but it's not good to ignore any
part of the system when choosing a record level.

There is no such thing as a "theoretical problem" where a human operator
will be involved. We start with an actual problem, the operator, the
parameters for which will be, as we say, all over the map. If you have
just imagined any problem derived either from design limitations or
erroneous settings made by a knob twister, that problem is no longer
theoretical. It can happen, therefore it will.

There is no substitute for a qualified operator. (May apply to
composers, too. ;-)

--
shut up and play your guitar * HankAlrich.Com
HankandShaidriMusic.Com
YouTube.Com/WalkinayMusic