[Mike Rivers]

On 11/20/2010 3:53 PM, rickman wrote:

I don't recall any such restriction on N. Certainly the measurement
has just as much meaning with N = 1 as any greater N. Think of it as
a limit as N approaches 1 then. The point is that it is as valid a
measurement for a single point as it is for many points. It
represents the equivalent voltage that would produce the same power as
DC of the same voltage.

Consider the effective value of RMS. It's the amount of heat
can be produced by the electrical energy. If you have only
one sample, no matter how high it is, do you really think it
will warm up your coffee?


--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Don Pearce[_3_]]

On Sun, 21 Nov 2010 07:26:52 -0500, Mike Rivers
wrote:

On 11/20/2010 3:53 PM, rickman wrote:

I don't recall any such restriction on N. Certainly the measurement
has just as much meaning with N = 1 as any greater N. Think of it as
a limit as N approaches 1 then. The point is that it is as valid a
measurement for a single point as it is for many points. It
represents the equivalent voltage that would produce the same power as
DC of the same voltage.

Consider the effective value of RMS. It's the amount of heat
can be produced by the electrical energy. If you have only
one sample, no matter how high it is, do you really think it
will warm up your coffee?

It will if you have a similarly tiny sample of coffee. I think you are
confusing power and energy here.

d

[Mike Rivers]

On 11/21/2010 5:22 AM, davew wrote:

This thread is about dBFS.

I don't think that it really is, though that's the only clue
we have from the header. Another clue was a screen shot of
some computer-generated bargraphs which were labeled "RMS."
I think that the troll is asking how it comes up with the
length of that bar. That's a legitimate question, but I
doubt that there's anyone here, with the possible exception
of Dick Pierce, who has ever been close enough to the design
of such a program to have an idea of what assumptions and
formula the programmer used.

Hence, the conclusion that, as an indicator of how loud the
audio sounds (relative to other chunks of audio, if you
don't adjust the volume control), it has some meaning. But
as a representation of a voltage, number of bits in use at a
particular time, or as an indication of how much headroom is
available, its value is only in the way that the user
interprets the bargraph reading and how it's changing over
time.

This is actually no different than the correct way to use a
VU meter.



--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Mike Rivers]

On 11/20/2010 6:16 PM, Ian Bell wrote:

dBFS is NOT a measurement method (peak or rms) but a
specification for a signal level. Unlike dBm, dBu and dBV is
has NO SPECIFIC PHYSICAL VALUE - it is simply the largest
value that a digital system can represent.

That's one way of looking at it. The other way of looking at
it is that when you know the relationship between full scale
and output level or input sensitivity, dBFS has a physical
meaning. However, it's more useful to specify a voltage (dBu
etc.) than dBFS when working with actual useful digital
audio hardware.

I've been trying to find out what the original poster's real
question is, but he seems to either not be sure or just
refuses to answer, rather enjoying saying "no, that's not
it" rather than formulate a question that isn't abstract.


--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Mike Rivers]

On 11/21/2010 2:01 AM, glen herrmannsfeldt wrote:

I sometimes record live high-school orchestra concerts.
Because it is hard to know the level, I record 24 bit, then
find the peak and RMS of each track. Then I figure out how many
bits to scale each track by so that peaks stay below FS, and
they should sound about right together.

Well, the idea is that they should sound about the same level,
such that one shouldn't want to run up and change the volume
control for each track.

Oh, you mean mixing in the sense of "mix tape" rather than
mixing a multitrack recording down to stereo. Well, of
course the listener shouldn't have to run up and change the
volume control, but, too, he shouldn't be subject to a flat
program unless he isn't actually listening to it. Elevator
music is good for that. Concerts, and even interesting radio
programming don't work that way, however.

If there's risk of hearing or speaker damage when switching
between songs in a program, sure, that should be fixed, and
normalizing to equal peak level can work. But unless each
piece of the program has very little dynamic range (which
isn't all that unusual in pop music today) you'll still have
differences in perceived loudness.

I don't think I could do that very
well just listening to them, trying to memorize the average
level over a 15 minute track.

That's not the way you do it. You consider how annoyed you
are when going from song to song. This is the money crop for
mastering engineers.




--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Ian Bell[_2_]]

Mike Rivers wrote:
On 11/20/2010 6:16 PM, Ian Bell wrote:

dBFS is NOT a measurement method (peak or rms) but a
specification for a signal level. Unlike dBm, dBu and dBV is
has NO SPECIFIC PHYSICAL VALUE - it is simply the largest
value that a digital system can represent.

That's one way of looking at it. The other way of looking at it is that
when you know the relationship between full scale and output level or
input sensitivity, dBFS has a physical meaning. However, it's more
useful to specify a voltage (dBu etc.) than dBFS when working with
actual useful digital audio hardware.

I've been trying to find out what the original poster's real question
is, but he seems to either not be sure or just refuses to answer, rather
enjoying saying "no, that's not it" rather than formulate a question
that isn't abstract.




He is also getting confused between the definition of signal level (like dBu for
example) and the means of measuring a signal level like RMS or peak.

Cheers

Ian

[Scott Dorsey]

Eric Jacobsen wrote:

So, in this "precise definition" how long does one integrate when
measuring a practical signal? Can one always expect results to be
independent of integration time?

You integrate for as long as the signal lasts. That is, if you have a
digital file, there is one and only one RMS average value for the whole
file.

If you keep a running average or a windowed average, what you are displaying
is not a standard RMS value.

Average-reading meters display running averages with different weighting
and averaging times. Consequently, standards for average-reading meters
have to list things like the time constants so the meter ballistics will
all be the same.

The VU meter spec is very specfic and it details precisely how the averaging
is to be done. Likewise the BS.1771 spec for digital averaging meters.

Most DAW average meters (some of which incorrectly call themselves RMS
meters) use arbitrary averaging and consequently they do not agree with
one another. You cannot say they are "dB-with-respect-to-something-specific"
at all. They are just useless flashing lights that provide little real
information until you know precisely how they are weighted.
--scott


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

[Mike Rivers]

On 11/21/2010 7:28 AM, Don Pearce wrote:

It will if you have a similarly tiny sample of coffee. I think you are
confusing power and energy here.

Think again, smart boy.


--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Eric Jacobsen]

On 21 Nov 2010 13:14:16 -0500, (Scott Dorsey) wrote:

Eric Jacobsen wrote:

So, in this "precise definition" how long does one integrate when
measuring a practical signal? Can one always expect results to be
independent of integration time?

You integrate for as long as the signal lasts. That is, if you have a
digital file, there is one and only one RMS average value for the whole
file.

What if it's a continuous stream of digital data from a data
collection or continuous control system? Can you not compute the
power level until it's shut off and the signal stops?

If you keep a running average or a windowed average, what you are displaying
is not a standard RMS value.

What is a "standard" RMS value? Which standard?

Average-reading meters display running averages with different weighting
and averaging times. Consequently, standards for average-reading meters
have to list things like the time constants so the meter ballistics will
all be the same.

Exactly. For uniformity there has to be further detail determined and
settled upon and communicated. Just saying "RMS average" is not
enough to convey a specific repeatable technique. It's close enough
in many cases, but not in some cases where the fine details matter.

The VU meter spec is very specfic and it details precisely how the averaging
is to be done. Likewise the BS.1771 spec for digital averaging meters.

That's one example of a detailed spec. It is not anything close to
universal.

Most DAW average meters (some of which incorrectly call themselves RMS
meters) use arbitrary averaging and consequently they do not agree with
one another. You cannot say they are "dB-with-respect-to-something-specific"
at all. They are just useless flashing lights that provide little real
information until you know precisely how they are weighted.
--scott


Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.abineau.com

[Scott Dorsey]

Eric Jacobsen wrote:
On 21 Nov 2010 13:14:16 -0500, (Scott Dorsey) wrote:

Eric Jacobsen wrote:

So, in this "precise definition" how long does one integrate when
measuring a practical signal? Can one always expect results to be
independent of integration time?

You integrate for as long as the signal lasts. That is, if you have a
digital file, there is one and only one RMS average value for the whole
file.

What if it's a continuous stream of digital data from a data
collection or continuous control system? Can you not compute the
power level until it's shut off and the signal stops?

Right. It's not a continuous measurement, it's a scalar.

If you keep a running average or a windowed average, what you are displaying
is not a standard RMS value.

What is a "standard" RMS value? Which standard?

There is only one standard. RMS is something very specific, with
a mathematical definition. Look in the ITT Radio Engineer's Handbook
if you don't believe me.

Average-reading meters display running averages with different weighting
and averaging times. Consequently, standards for average-reading meters
have to list things like the time constants so the meter ballistics will
all be the same.

Exactly. For uniformity there has to be further detail determined and
settled upon and communicated. Just saying "RMS average" is not
enough to convey a specific repeatable technique. It's close enough
in many cases, but not in some cases where the fine details matter.

RMS average is a specific and repeatable technique and it produces a
single scalar value from a data stream.

There are a lot of manufacturers who call various moving average systems
"RMS." You can call your dog a cat, too. But that doesn't make him a
cat. And he won't eat the cat food.

The VU meter spec is very specfic and it details precisely how the averaging
is to be done. Likewise the BS.1771 spec for digital averaging meters.

That's one example of a detailed spec. It is not anything close to
universal.

It's not universal at all. Hardly anyone uses it. This is the problem.

Most DAW average meters (some of which incorrectly call themselves RMS
meters) use arbitrary averaging and consequently they do not agree with
one another. You cannot say they are "dB-with-respect-to-something-specific"
at all. They are just useless flashing lights that provide little real
information until you know precisely how they are weighted.

I just want to reiterate this again. Unless it is built to a known spec,
unless it says "LUFS" on the averaging digital meter, or "VU" on the
averaging analogue meter, then it's not a real measurement, it is only
usable for qualitative estimation, and it is not to be counted on.
--scott


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

[Adrian Tuddenham[_2_]]

Scott Dorsey wrote:

[...]
... You can call your dog a cat, too. But that doesn't make him a
cat. And he won't eat the cat food.

He will if he thinks the cat wants it.

--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

[Mike Rivers]

On 11/21/2010 3:11 PM, Eric Jacobsen wrote:

there are an infinite number of different ways to calculate RMS, and
unless it's specified, you won't know what was done.

Oh, no, there's only one way to CALCULATE the RMS value -
the square root of the sum of the squares. But there are
many ways to MEASURE the values from which you're
calculating it. Most significant is how the voltage varies
with time and how many measurements to you include in the
calculation over how much time.

If the program performs the calculation correctly, you can
get good results from a DAW that's capable of displaying RMS
amplitude over whatever length of time you choose - an
entire file, a selected portion of it, even just a couple of
cycles. It has plenty of samples to work from. In this case,
the RMS amplitude is almost always expressed as dBFS because
that's what the program knows.

If it's an analog measurement, you read the meter for as
long as you want to watch it, take an eyeball average if the
waveform isn't constant, and use that number for whatever
you wish. As has been discussed before, many voltmeters are
calibrated assuming that you're measuring a sine wave. True
RMS meters actually do some calculations. Some use
integrators. some even use thermocouples.


--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Don Pearce[_3_]]

On Sun, 21 Nov 2010 13:45:37 -0500, Mike Rivers
wrote:

On 11/21/2010 7:28 AM, Don Pearce wrote:

It will if you have a similarly tiny sample of coffee. I think you are
confusing power and energy here.

Think again, smart boy.

Thought again - just in case. No change.

d

[glen herrmannsfeldt]

In comp.dsp Mike Rivers wrote:
(snip, I wrote)

(snip, and previous snip, on record levels)

Well, the idea is that they should sound about the same level,
such that one shouldn't want to run up and change the volume
control for each track.

Oh, you mean mixing in the sense of "mix tape" rather than
mixing a multitrack recording down to stereo. Well, of
course the listener shouldn't have to run up and change the
volume control, but, too, he shouldn't be subject to a flat
program unless he isn't actually listening to it. Elevator
music is good for that. Concerts, and even interesting radio
programming don't work that way, however.

Yes. The level I am considering is per track. One that I have
done had some tracks a symphony orchestra and others a flute
quintet, three harps, or sometimes just vocals.

If there's risk of hearing or speaker damage when switching
between songs in a program, sure, that should be fixed, and
normalizing to equal peak level can work. But unless each
piece of the program has very little dynamic range (which
isn't all that unusual in pop music today) you'll still have
differences in perceived loudness.

Yes, so I usually don't want to go for just peaks. It isn't
(usually) pop, but vocal music has much less dynamic range than
the symphony orchestra.

(snip)

That's not the way you do it. You consider how annoyed you
are when going from song to song. This is the money crop for
mastering engineers.

There is no money in this, though.

-- glen

[Randy Yates]

(Scott Dorsey) writes:
[...]
I just want to reiterate this again. Unless it is built to a known spec,
unless it says "LUFS" on the averaging digital meter, or "VU" on the
averaging analogue meter, then it's not a real measurement, it is only
usable for qualitative estimation, and it is not to be counted on.

That's a pretty sad state of affairs for pro audio, don't you think
Scott?
--
Randy Yates % "Midnight, on the water...
Digital Signal Labs % I saw... the ocean's daughter."
% 'Can't Get It Out Of My Head'
http://www.digitalsignallabs.com % *El Dorado*, Electric Light Orchestra

[Mark]

The point being, then, that there is not a single, precise definition.
In the discrete case n can change and in the continuous case the
integration time can change. * As Robert said quite a while back,
there are an infinite number of different ways to calculate RMS, and
unless it's specified, you won't know what was done.

I suspect this was the source of Randy's initial question regarding
dBFS, since it is pretty ambiguous.

Not really. At least that wasn't the most important question in my
mind.

The question in my mind is this:

* 1. How is dBFS defined? If there is no formal definition, then
* how is it used?

Specifically, what I was wondering was this:

* 1. Is dBFS an instantaneous measurement or some sort of average?

* 2. If dBFS is some sort of average, what type? RMS? RMS sine? RMS
* square? Averaged magnitude? Any of an infinite number of ways to
* define it?

I admit there is some ambiguity built into the question itself since, in
my understanding and teaching, a dB (of WHATEVER) can always be traced
back to a ratio of two powers, and power, as I understand it, is always
an average [Note 1]. Note that this "traceback path" to power can be constructed
even for the digital dBFS by assuming a conversion factor (x Volts /
units) and a load impedance (probably 1 ohm). But I am allowing for the
possibility of an "instantaneous" definition/usage.

Also, let me acknowledge that many folks have already answered my
question. But just to clarify YOUR (implied) question, Eric, (which is,
"What was Randy's initial question?") and possibly others that still
don't understand what and why I'm asking, here it is.

--Randy

[1] I acknowledge that, as others have mentioned, I could be wrong here,
but I currently remain unconvinced.
--
Randy Yates * * * * * * * * * * *% "She tells me that she l


dB is also sometimes used for Voltage ratios as in dBmV and in this
case it is RMS Voltage so we have come full circle... :-)

Standard definitions for RMS apply only to time invariant waveforms
like steady tones or steady square waves etc.

There is no standard for RMS for time varying signals ...

Mark

[Randy Yates]

Mark writes:

The point being, then, that there is not a single, precise definition.
In the discrete case n can change and in the continuous case the
integration time can change. Â* As Robert said quite a while back,
there are an infinite number of different ways to calculate RMS, and
unless it's specified, you won't know what was done.

I suspect this was the source of Randy's initial question regarding
dBFS, since it is pretty ambiguous.

Not really. At least that wasn't the most important question in my
mind.

The question in my mind is this:

Â* 1. How is dBFS defined? If there is no formal definition, then
Â* how is it used?

Specifically, what I was wondering was this:

Â* 1. Is dBFS an instantaneous measurement or some sort of average?

Â* 2. If dBFS is some sort of average, what type? RMS? RMS sine? RMS
Â* square? Averaged magnitude? Any of an infinite number of ways to
Â* define it?

I admit there is some ambiguity built into the question itself since, in
my understanding and teaching, a dB (of WHATEVER) can always be traced
back to a ratio of two powers, and power, as I understand it, is always
an average [Note 1]. Note that this "traceback path" to power can be constructed
even for the digital dBFS by assuming a conversion factor (x Volts /
units) and a load impedance (probably 1 ohm). But I am allowing for the
possibility of an "instantaneous" definition/usage.

Also, let me acknowledge that many folks have already answered my
question. But just to clarify YOUR (implied) question, Eric, (which is,
"What was Randy's initial question?") and possibly others that still
don't understand what and why I'm asking, here it is.

--Randy

[1] I acknowledge that, as others have mentioned, I could be wrong here,
but I currently remain unconvinced.
--
Randy Yates Â* Â* Â* Â* Â* Â* Â* Â* Â* Â* Â*% "She tells me that she l


dB is also sometimes used for Voltage ratios as in dBmV and in this
case it is RMS Voltage so we have come full circle... :-)

dbV is also another. However, both of these can be traced back to power
with some assumptions. Also, as has been pointed out, if these types
of measurements were only voltage, why was the multiplier changed from
10 to 20? Surely we must agree that the prefix "deci" means by ten!

Standard definitions for RMS apply only to time invariant waveforms
like steady tones or steady square waves etc.

There is no standard for RMS for time varying signals ...

I understand that there will be some ambiguity for such measurements.
That's not my primary concern. My primary concern is what the target
bogie is: (the theoretical ideal) RMS or whatever else.
--
Randy Yates % "My Shangri-la has gone away, fading like
Digital Signal Labs % the Beatles on 'Hey Jude'"
%
http://www.digitalsignallabs.com % 'Shangri-La', *A New World Record*, ELO

[Eric Jacobsen]

On 21 Nov 2010 15:44:14 -0500, (Scott Dorsey) wrote:

Eric Jacobsen wrote:
On 21 Nov 2010 13:14:16 -0500, (Scott Dorsey) wrote:

Eric Jacobsen wrote:

So, in this "precise definition" how long does one integrate when
measuring a practical signal? Can one always expect results to be
independent of integration time?

You integrate for as long as the signal lasts. That is, if you have a
digital file, there is one and only one RMS average value for the whole
file.

What if it's a continuous stream of digital data from a data
collection or continuous control system? Can you not compute the
power level until it's shut off and the signal stops?

Right. It's not a continuous measurement, it's a scalar.

The point was that often a power measurement is needed in real time
for a continuing signal. You can't wait until the system stops to
get a measurement. n can always be a subset of N.

If you keep a running average or a windowed average, what you are displaying
is not a standard RMS value.

What is a "standard" RMS value? Which standard?

There is only one standard. RMS is something very specific, with
a mathematical definition. Look in the ITT Radio Engineer's Handbook
if you don't believe me.

No, there are tons of standards. Some agree, some don't. That's the
point. The mathematical defnitions of RMS are well know, and none
(that are worth looking at) define the length of time used for
integration of a non-periodic signal with anything but a variable.
i.e., the integration time is not part of the "definition". This is
part of the point I'm making.

Average-reading meters display running averages with different weighting
and averaging times. Consequently, standards for average-reading meters
have to list things like the time constants so the meter ballistics will
all be the same.

Exactly. For uniformity there has to be further detail determined and
settled upon and communicated. Just saying "RMS average" is not
enough to convey a specific repeatable technique. It's close enough
in many cases, but not in some cases where the fine details matter.

RMS average is a specific and repeatable technique and it produces a
single scalar value from a data stream.

For what integration time?

There are a lot of manufacturers who call various moving average systems
"RMS." You can call your dog a cat, too. But that doesn't make him a
cat. And he won't eat the cat food.

I like cats.

The VU meter spec is very specfic and it details precisely how the averaging
is to be done. Likewise the BS.1771 spec for digital averaging meters.

That's one example of a detailed spec. It is not anything close to
universal.

It's not universal at all. Hardly anyone uses it. This is the problem.

Most DAW average meters (some of which incorrectly call themselves RMS
meters) use arbitrary averaging and consequently they do not agree with
one another. You cannot say they are "dB-with-respect-to-something-specific"
at all. They are just useless flashing lights that provide little real
information until you know precisely how they are weighted.

I just want to reiterate this again. Unless it is built to a known spec,
unless it says "LUFS" on the averaging digital meter, or "VU" on the
averaging analogue meter, then it's not a real measurement, it is only
usable for qualitative estimation, and it is not to be counted on.
--scott

I disagree.


Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.abineau.com

[Mr.T]

"Dick Pierce" wrote in message
...
glen herrmannsfeldt wrote:
In comp.dsp Dick Pierce wrote:

(snip, I wrote)


In EE, RMS can mean different things.

Not in any "EE" I ever encountered. It means one thing:

sqrt(sum(X1^2 + X2^2 ... +Xn^2) / n)

It used to be, for DVMs, that was called True RMS. When
they figured out how to actually do that electrically.

Maybe they all do now, so it isn't a problem anymore,
but it didn't used by be that way. Also, it isn't
for the peak-to-peak reading VTVM, normally calibrated
with an RMS (assuming sine) scale.

Sorry, wrong. RMS, or "root-mean-square," is a term
that predates voltmeters and DVMs and VTVMs. It is a
mathemateical, yea, a statistical mathematics defintion.

That is has been misused, misappropriated, improperly
implemented, misunderstood or whatever does NOT change
the fact that it has a precise, well understood and
universally agreed-upon mathematical definition.

That you can cite one or more examples of a piece of
instrumentation that claims "RMS" merely points out
that there are different claims, some valid, some not,
of implementations. It does NOT siggest that there are
different meanings to the term.


He's wrong in any case, ANY meter which was average responding and
calibrated in RMS, ALWAYS assumed a SINE wave input and was corrected for
it. That's why True RMS meters were necessary for anything other than pure
sine waves.

MrT.

[Mr.T]

"Randy Yates" wrote in message
...
dbV is also another. However, both of these can be traced back to power
with some assumptions.

Not assumptions, definitions, like a fixed impedance.

Also, as has been pointed out, if these types
of measurements were only voltage, why was the multiplier changed from
10 to 20?

Didn't you cover P=E*I and it's application to voltage given a fixed
impedance in your studies?
P=E^2/R and when using the log system for dB you multiply the 10*2 to get 20
(same for current since P=I^2R)

Surely we must agree that the prefix "deci" means by ten!

By 1/10th actually, and YES it's ALWAYS 1/10th of a BELL!
You are simply allowing for the E^2 term when using 20 instead of 10 for
voltage.

MrT.

[Peter Larsen[_3_]]

glen herrmannsfeldt wrote:

Yes, so I usually don't want to go for just peaks. It isn't
(usually) pop, but vocal music has much less dynamic range than
the symphony orchestra.

Not before the recording process, but perhaps after.

-- glen

Kind regards

Peter Larsen

[Arny Krueger]

"davew" wrote in message


The VU meter is basically a bridge rectifier followed by
a low pass filter. So it's mean rectified, not mean squared.



So a 1dB difference for pure tone. We don't tend to use rms or
mean whatever when talking about audio levels though, we just say
"level" and that seems to be good enough. It's understood that when you
reach 0dBFS you're in trouble shortly thereafter.

http://www.allaboutcircuits.com/vol_2/chpt_1/3.html

" For RMS measurements, analog meter movements (D'Arsonval, Weston, iron
vane, electrodynamometer) will work so long as they have been calibrated in
RMS figures. Because the mechanical inertia and dampening effects of an
electromechanical meter movement makes the deflection of the needle
naturally proportional to the average value of the AC, not the true RMS
value, analog meters must be specifically calibrated (or mis-calibrated,
depending on how you look at it) to indicate voltage or current in RMS
units. The accuracy of this calibration depends on an assumed waveshape,
usually a sine wave. "

IOW, ordinary passive meters are not True RMS meters without a lot of help.

The averaging function is produced mechanically, by the mass of the moving
parts of the meter

If you want to make a passive analog meter that properly reads True RMS, it
must use a movement that responds to the square of the applied voltage, and
is nonlinearly calibrated to indicate the square root of the applied
voltage.

A meter that uses the applied voltage to establish both the magnetic field
around its armature, and also the magnetic field produced by the armature;
will respond to the square of the applied voltage. With this kind of meter,
doubling both voltages (which would ordinarly be based on the same signal)
will interact to produce an attraction or repulsion that quadruples. The
pointer will deflect 4 times further, and the nonlinear calibration of the
scale will provide the square root function.

Most meters use a permanent magnet to establish the magnetic field around
the armature because this makes it easier to produce a sensitive meter. They
will then respond more or less linearly to the intensity of the applied
voltage. Therefore a proper True RMS meter will probably not be the most
sensitive meter around.

Passive True RMS meters were widely used in the days before complex active
electronic circuits became a practical means to accomplish the same end. I
have one that was sold by RCA back in the day for measuring power line
voltage, which has the expected highly nonlinear scale.

[Arny Krueger]

"Randy Yates" wrote in message

(Scott Dorsey) writes:
[...]
I just want to reiterate this again. Unless it is built
to a known spec, unless it says "LUFS" on the averaging
digital meter, or "VU" on the averaging analogue meter,
then it's not a real measurement, it is only usable for
qualitative estimation, and it is not to be counted on.

That's a pretty sad state of affairs for pro audio, don't
you think Scott?

Not at all. Pro audio is still heavily dependent on tools that do not
totally perform in accordance with simple neat, precise theories.

That's one reason why most of us consider audio to be both a science and an
art. The science is in the things that do perform in accordance with simple,
neat, precise theories which abound but does not include everything. The
art is in how we manage the many tools and practices that do not totally
perform in accordance with simple neat, precise theories, as used. Things
like microphones and loudspeakers.

[Arny Krueger]

"Randy Yates" wrote in message


What units would a typical professional digital audio
system use to measure RMS values of digital signals?

The well known DAW software packages Cool Edit Pro and Adobe Audition both
have a measurement tool that produces both peak, average, and RMS values.
The measurement interval is the range of the wave that is selected and
highlighted on the screen. They provide their measurements in dB FS.

Signal values in DAW programs are typically dimensionless because they are
dependent on the gain of the ADC that was used to digitize them, perchance
they once existed as analog voltages. The gain of ADCs is not standardized
and is often not even carefully specified. Pro audio ADCs often have
continuously-variable analog or digital attenuators on their inputs. If you
stabilize the settings of the input attenuators, then Pro Audio ADCs can be
calibrated. I use an analog meter and sine waves for the purpose.

The RMS and peak values of a dimensionless quantity are themselves both
dimensionless.

[Arny Krueger]

"Dick Pierce" wrote in message

Scott Dorsey wrote:
I don't know, because I have never seen actual
instantaneous RMS values ever displayed anywhere.

The term "instantaneous RMS value" is itself meaningless.

Right, because RMS has to be averaged over an interval.

Instantaneous peak values can be meaningful, but in the digital domain they
aren't really instantaneous. They correpond to a sample interval.

As a practical matter, some measuring tools allow the measurement interval
can be set to be within a useful range that corresponds to something of
interest to the user.

[Arny Krueger]

"Mike Rivers" wrote in message

On 11/20/2010 3:53 PM, rickman wrote:

I don't recall any such restriction on N. Certainly the
measurement has just as much meaning with N = 1 as any
greater N. Think of it as a limit as N approaches 1
then. The point is that it is as valid a measurement
for a single point as it is for many points. It
represents the equivalent voltage that would produce the
same power as DC of the same voltage.

Consider the effective value of RMS. It's the amount of
heat can be produced by the electrical energy. If you
have only one sample, no matter how high it is, do you
really think it will warm up your coffee?

Depends on the sample rate. If the sample rate is one sample every 10
minutes... ;-)

[Mark]

some comments:

1) Any meter can be calibrated to read the RMS value of a steady sine
wave. i.e. Simpson 260 reads the RMS value of a steady sine wave but
is not true RMS responding.

2) True RMS meters will read the RMS value of any (within its
limitation) STEADY waveform i.e. a steady square wave or steady
triangle wave etc.

3) There is no standardized reading of RMS for a time varying waveform
like real audio, as has been mentioned the integration time and
weighting needs to be defined and I don't think there is such a
standard?

4) dBFS meters on most digital equipment are not RMS or average but
are more peak reading, that (try to) capture the peak value of even an
individual sample.

5) Most of us given a choice I suspect would choose to a dual meter
system, one that shows both the peak so that we can be assured there
is no clipping combined with another display that show the LOUDNESS.
LOUDNESS is not the same as RMS. There are some standards on metering
of loudness of audio.

Those are the two reasons we meter audio, to assure the equipment is
not overloaded even on peaks and for some gauge of the loudness.

Note: As has been discussed in another thread on intersampling peaks,
even the definition of peak is ambiguous, the peak SAMPLE value is not
always the same as the peak value of the reconstructed waveform. I
don't know, but I suspect that most peak meters calibrated in dBFS are
reading the peak SAMPLE value and not the true wavefomr peak therfore
the true audio peak can be even higher. I don't know if this is what
Randy is trying to get at or not.

Mark

[Mike Rivers]

On 11/21/2010 8:54 PM, Randy Yates wrote:

That's a pretty sad state of affairs for pro audio, don't you think
Scott?

"Pro Audio" doesn't call the shots. It's the suppliers of
program material to the consumers that do. They're the ones
who can do things that earn them more money. All the
engineer can do is either follow instructions of those
paying him and stay in business, or follow his heart,
conscience, and good taste and get a little less, or less
profitable, work.

--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[davew]

On Nov 22, 1:51*pm, Mike Rivers wrote:
On 11/21/2010 8:54 PM, Randy Yates wrote:

That's a pretty sad state of affairs for pro audio, don't you think
Scott?

"Pro Audio" doesn't call the shots. It's the suppliers of
program material to the consumers that do. They're the ones
who can do things that earn them more money. All the
engineer can do is either follow instructions of those
paying him and stay in business, or follow his heart,
conscience, and good taste and get a little less, or less
profitable, work.

--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com- useful and
interesting audio stuff

Pro-audio is beginning to call the shot's though. Under pressure from
the consumer who is fed up of turning the volume down in ad breaks.
Broadcasters will soon be required to stay within certain limits of
the new loudness scale. I doubt it will have much of an immediate
impact on recorded music though it will still apply when broadcast so
it may actually indirectly result in better quality music production
in the long run (we all hope).

[Mike Rivers]

On 11/21/2010 9:02 PM, Randy Yates wrote:

The question in my mind is this:

1. How is dBFS defined? If there is no formal definition, then
how is it used?

You seem to think that the formal definition - that it's a
SCALE and not an absolute value, is inadequate, but that's
all there is. Sorry.

It's like asking how Fahrenheit is defined. You can define
212 degrees F as the temperature at which water boils (under
specified conditions), but you can't define what "a
Fahrenheit" is.

You've also been explained how dBFS is used. Once again,
with feeling, any value below zero tells you how much
headroom you have. That's all. It doesn't tell you how loud
something sounds unless you add time and duty cycle (which
will tell you how much air gets moved at the end of the
chain), and you can't use it to determine how many volts
will come out the analog end or how many volts coming in
will get you to a specific number on the dBFS scale.

Specifically, what I was wondering was this:

1. Is dBFS an instantaneous measurement or some sort of average?

The only significant number is instantaneous. You could
calculate an average when you know all the values within the
bounds over which you're calculating it. Although some
programs do that (and give you a number) because they CAN,
it really isn't of general use. I think that's why everyone
is dancing around an answer for you. We don't use "average
dBFS" so we don't care. There are other, more useful averages.

2. If dBFS is some sort of average, what type? RMS? RMS sine? RMS
square? Averaged magnitude? Any of an infinite number of ways to
define it?

Will you accept "no" as a temporary answer until you
yourself define how you'd like it defined, and for what purpose?

I admit there is some ambiguity built into the question itself since, in
my understanding and teaching, a dB (of WHATEVER) can always be traced
back to a ratio of two powers, and power, as I understand it, is always
an average [Note 1].

Classically, dB is a ratio of measured power to a reference
power. Mathematically, however, it's a ratio of anything you
want. It doesn't have to be power, and it can be an
instantaneous measurement. In the case where we use dBFS,
the reference is 0 dBFS, and the amplitude of any single
sample, as instantaneous as you can get, can be compared to
the reference for a single value of dBFS.

We can, in fact, even do better than that. Under certain
conditions, with or without less than theoretically perfect
filtering, it's possible to calculate a value greater than 0
dBFS given data from two or more adjacent samples. This is
NOT an average, however. And it's real. It's a source of
distortion in A/D conversion.

Also, let me acknowledge that many folks have already answered my
question.

Can we give it a rest, then? What does Hitler have to say
about dBFS?

--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Arny Krueger]

"Mark" wrote in message

some comments:

1) Any meter can be calibrated to read the RMS value of a
steady sine wave. i.e. Simpson 260 reads the RMS value of
a steady sine wave but is not true RMS responding.

Yes, and most of the meters in use, whether as test equipment or part of a
piece of audio gear, is average-responding, calibrated as if it was RMS.
IOW as we both knows, its only accurate for sine waves.

2) True RMS meters will read the RMS value of any (within
its limitation) STEADY waveform i.e. a steady square wave
or steady triangle wave etc.

That depends on the TRMS meter. In another post I pointed out that some DAW
software metering facilities respond to the portion of the wave that is
selected which can range from a whole song to one or just a few samples.

3) There is no standardized reading of RMS for a time
varying waveform like real audio, as has been mentioned
the integration time and weighting needs to be defined
and I don't think there is such a standard?

I'm pretty comfortable with the metering in my DAW software because it
leaves the integration time up to me and delivers both peak and average
readings.

4) dBFS meters on most digital equipment are not RMS or
average but are more peak reading, that (try to) capture
the peak value of even an individual sample.

Yes, or they respond to both peak and average values.

5) Most of us given a choice I suspect would choose to a
dual meter system, one that shows both the peak so that
we can be assured there is no clipping combined with
another display that show the LOUDNESS. LOUDNESS is not
the same as RMS. There are some standards on metering of
loudness of audio.

I have a lot of equipment and software that works this way.

Those are the two reasons we meter audio, to assure the
equipment is not overloaded even on peaks and for some
gauge of the loudness.

Agreed. Our ears response is closely approximated by True RMS as
calculated over various periods of time.

Note: As has been discussed in another thread on
intersampling peaks, even the definition of peak is
ambiguous, the peak SAMPLE value is not always the same
as the peak value of the reconstructed waveform.

I've been aware of this issue for years. I take intersample peaks to be
freaks of nature that respond well to being ignored.

I don't
know, but I suspect that most peak meters calibrated in
dBFS are reading the peak SAMPLE value and not the true
wavefomr peak therfore the true audio peak can be even
higher. I don't know if this is what Randy is trying to
get at or not.

Some software applies a Sinc function to the samples and integrates under
it.

[Mike Rivers]

On 11/22/2010 8:56 AM, davew wrote:

Pro-audio is beginning to call the shot's though. Under pressure from
the consumer who is fed up of turning the volume down in ad breaks.
Broadcasters will soon be required to stay within certain limits of
the new loudness scale.

How is that "Pro audio calling the shots?" It's the law,
broadcasters must comply. Pro Audio is giving the tools to
make the measurements (and arguing about it).

There have always been mastering engineers (I'd dare say the
majority) who support the concepts of dynamic range and
crest factors and in general their clients concur that this
sort of production sounds best. BUT when it comes to not
losing sales, many opt for "make it loud."

--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Don Pearce[_3_]]

On Mon, 22 Nov 2010 10:58:30 -0500, Dick Pierce
wrote:

Since energy is power integrated over an interval of time,
we have all the information we need. The power is whatever
the scaled sample value is. The interval is 1 sample period.
Thus the energy is knowable.

The sample period is the interval between samples, not the duration
over which a sample is integrated. As far as I know, all ADCs sample
an instant as the edge goes low. Energy at an instant is always zero,
whereas power at an instant is whatever it is.

d

[Don Pearce[_3_]]

On Mon, 22 Nov 2010 11:58:45 -0500, Dick Pierce
wrote:

Don Pearce wrote:
On Mon, 22 Nov 2010 10:58:30 -0500, Dick Pierce
wrote:


Since energy is power integrated over an interval of time,
we have all the information we need. The power is whatever
the scaled sample value is. The interval is 1 sample period.
Thus the energy is knowable.


The sample period is the interval between samples, not the duration
over which a sample is integrated. As far as I know, all ADCs sample
an instant as the edge goes low.

No, for simple implementation reasons, the aperture,
while small, is not zero. It is much smaller than
the sample interval, but the fact that is is non-zero
leads to a seldom-quoted A/D spec called "aperture
error". But, as I mention elswhere, sampling theorom
suggests that the difference between an aperture of 0
and an aperture approaching a sample width is
irrelevant to the resulting data.

This is all true, I accept that you have trumped my nit-pick with an
even smaller nit-pick (and so ad infinitum)

Energy at an instant is always zero,

Mr. Einstein muight disagree, given that mass
can exist during that same instant.

Things approaching zero gave Mr Einstein considerable trouble. He
wasn't at all happy about quantum mechanics.

whereas power at an instant is whatever it is.

But, as you will probably agree, there are no "instants"
in any practical system we can build.

No, no instants - but we do go an awful lot smaller than the sampling
interval.

d

[Scott Dorsey]

In article , Randy Yates wrote:
(Scott Dorsey) writes:
[...]
I just want to reiterate this again. Unless it is built to a known spec,
unless it says "LUFS" on the averaging digital meter, or "VU" on the
averaging analogue meter, then it's not a real measurement, it is only
usable for qualitative estimation, and it is not to be counted on.

That's a pretty sad state of affairs for pro audio, don't you think
Scott?

Not really, since people very seldom really need to know average levels
these days. What is important in the digital world are peak levels and
perceived loudness... and dBFS takes care of the first one nicely.

If you really need to know average levels for some reason, you can buy
calibrated meters from Dorrough or RTW which follow the ITU standard.
--scott

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

[Ben Bradley[_2_]]

On Fri, 19 Nov 2010 16:33:53 -0500, Randy Yates
wrote:

Hi,

Some had responded here to my recent inquiry on levels that dBFS is a
peak measurement.

If an RMS measurement needs to be made for a digital signal (i.e., on a
digital mixing console or a ProTools plugin), what units are utilized? I
thought they were dBFS, i.e., that dBFS was an RMS measurement.
Apparently I am incorrect. Somebody please set me straight.

What an interesting thread. It's funny the things that get argued
again and again ... good to see that Usenet is still alive!

My answer is that dBFS is a peak measurement (and most likely of
samples, not of a reconstructed waveform, which could be higher) as
displayed by the bargraph "meters" in most recording software, unless
it says it's something else. The something else is usually RMS.

There's been debate in this thread about bits vs. voltages (or
power, which is the "root" (sorry) of the RMS voltage measurement). I
really see no conflict here. The dB levels calculated with bits (more
specifically, linear PCM representations of discrete sampled voltages)
are the same as when the samples are put out through DAC's and the
relative voltages measured and the dB values calculated from those.
The numeric values are just binary representations of voltages, and
the values are calculated the same way. The dB value is the ratio of
two values of power, or two values of RMS voltage into a fixed
resistor, or between two sets of samples whose RMS values are
calculated.

This reminds me of a thread I read here (rec.audio.pro) long ago on
dBFS RMS measurement, and how one recording program did it.

First is a setup thread, "The Crest of the Wave (reference stuff!):"

http://groups.google.com/group/rec.a...213daf805e93c9

Next, the thread I remember: "RMS in CEP, just in from Syntrillium"
(that's Cool Edit Pro, presumably used the same RMS code as Cool Edit
96, both by Syntrillium before Adobe bought it all and renamed it
Audition):

http://groups.google.com/group/alt.a...cacf7552df88d7

I hope that sheds more light than heat. It's also notable that
there are several participants in those 11-year-old threads that are
also posting in this one.

[Michael Dines[_2_]]

Scott Dorsey wrote:

For example, the average meters on Pro Tools don't seem to match anything
else or meet any known standard. The ballistics are faster than VU.

This answers something I've long wondered about - what do the meters on
the Pro Tools mix window measure when recording or playing back?

It's like the markings on the side of the waveform in the edit window -
are they meant to mean anything in dB? I've never foud anything in the
Pro Tools manuals to say what they're meant to mean.

[Mike Rivers]

On 11/23/2010 4:13 PM, Michael Dines wrote:

This answers something I've long wondered about - what do the meters on
the Pro Tools mix window measure when recording or playing back?

That there's audio there. They also give you an idea of how
close you're getting to clipping.

It's like the markings on the side of the waveform in the edit window -
are they meant to mean anything in dB? I've never foud anything in the
Pro Tools manuals to say what they're meant to mean.

I don't know about Pro Tools, but it's typical for waveform
graphics in a DAW to be scaled in dB, with the top and
bottom of the graph area representing full scale positive
and negative. This is the reason why everybody thinks his
mixes aren't "hot enough." With waveform peaks reaching a
fairly respectable -6 dBFS, the waveform fills only half the
area and looks pretty wimpy.

--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Bill Graham]

Mike Rivers wrote:
On 11/21/2010 8:54 PM, Randy Yates wrote:

That's a pretty sad state of affairs for pro audio, don't you think
Scott?

"Pro Audio" doesn't call the shots. It's the suppliers of
program material to the consumers that do. They're the ones
who can do things that earn them more money. All the
engineer can do is either follow instructions of those
paying him and stay in business, or follow his heart,
conscience, and good taste and get a little less, or less
profitable, work.

Not only that, but all the advancements in audio engineering have been made
for consumer equipment. If the big name manufacturers had to survive on what
they make and sell to the pros, they would have gone out of business a long
time ago. So, don't brush off us consumers as a thorn in your sides. If it
weren't for us, you would be out of business.....

[Mr.T]

Mike Rivers wrote:
"Pro Audio" doesn't call the shots. It's the suppliers of
program material to the consumers that do. They're the ones
who can do things that earn them more money. All the
engineer can do is either follow instructions of those
paying him and stay in business, or follow his heart,
conscience, and good taste and get a little less, or less
profitable, work.

Yep, if you want a good example of how sad it has become, grab the new Katy
Perry song "California Gurls" and open it into your wave editor of choice,
and have a look at just how badly compressed, clipped and grossly distorted
the waveforms are for a million selling record these days. It's nothing new
of course, the same was being done with Madonna, Brittany Spears etc.
However I didn't think it could get worse, and yet it has. As the
performance level of all the studio equipment used has increased, the
quality of the finished recordings has markedly decreased. :-(
It's not all due to incompetence of course, but unfortunately a lot of it is
if they can't get a similar result without so much distortion. I know many
recording/mastering engineers who can!

MrT.