I'm confused! White Noise Pink Noise A-weighted C-weighted
ChrisCoaster
is "flat".
But which one is "pink" and which is "white"??
For once and for all set me straight on the terminology.
-CC
Dave C
"ChrisCoaster" <ckoz...@snet.net> wrote in message
news:15a95a88-3336-400b...@x41g2000yqd.googlegroups.com...
Pink noise is "shaped" to achieve equal power / octave. White is "flat"
resulting in increasing power / octave.
Dave
Soundhaspriority
"ChrisCoaster" <ckoz...@snet.net> wrote in message
news:15a95a88-3336-400b...@x41g2000yqd.googlegroups.com...
There is a graph at http://en.wikipedia.org/wiki/A-weighting_curve
None of these are white, though they are all approximately white at the
left, low end of the graph. True white noise is a mathematical curiosity. It
is defined as equal power graphed against linear frequency. Since frequency
goes up forever, a true white noise signal would have infinite power. So all
the graphs convert to something other than linear as the frequency goes up.
As soon as the curve goes horizontal against a log scale of frequency, it is
not mathematically white.
Now look at "C". "C" is not linear per frequency, but it is linear per
octave: same power in each octave, between 100 and 3000 Hz. This is not
white noise, but someone decided it makes good practical sense to define
such a curve as flat.
Bob Morein
(310) 237-6511
William Sommerwerck
news:15a95a88-3336-400b...@x41g2000yqd.googlegroups.com...
> I know that A-weighting approximates human sensitivity
> and C-weighting is "flat".
> But which one is "pink" and which is "white"?
> For once and for all set me straight on the terminology.
You are confused, and the answers already posted are, in varying degrees,
incomplete/incorrect.
The weighting curves -- A and C -- have nothing to do with color. As you
say, C is basically flat, while A approximates the sensitivity of the human
ear at low levels.
"White" and "pink" refer to the noise test signals. White noise has equal
energy per bandwidth. For example, the 8ve between 1000 and 2000Hz has four
times the bandwidth and thus four times the energy as the 8ve between 250
and 500Hz. Thus, the energy rises at 3dB per 8ve. This is not what we think
of as "flat", though there is nothing "wrong" about white noise. That's just
the way it is.
Pink noise is white noise that's been run through a "pinking" filter with a
3dB/8ve rolloff. It has equal energy per octave. On an analyzer with 8ve or
third-8ve filters, the analyzer displays a "flat" response to pink noise.
This makes it easier to interpret the display, because it fits the way we
think about response.
Scott Dorsey
>I know that A-weighting approximates human sensitivity and C-weighting
>is "flat".
Not really. C-weighting is very far from flat.
>But which one is "pink" and which is "white"??
Neither one. The "pink" and "white" curves are used for noise sources,
the A-weighting and C-weighting for measuring devices. They are unrelated
curves and not used for the same things.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Richard Webb
>>But which one is "pink" and which is "white"??
> Neither one. The "pink" and "white" curves are used for noise
> sources, the A-weighting and C-weighting for measuring devices.
> They are unrelated curves and not used for the same things.
INdeed. WOuld you care to expound on when one might use a
weighted and when C weighted, since that would be his next
question, and would probably be beneficial. IN fact, if we
were still updating the faq that might be a good addition.
Regards,
Richard
--
| Remove .my.foot for email
| via Waldo's Place USA Fidonet<->Internet Gateway Site
| Standard disclaimer: The views of this user are strictly his own.
William Sommerwerck
> weighted and when C weighted, since that would be his next
> were still updating the FAQ that might be a good addition.
As I and others said, the A weighting mimics the response of the ear at low
levels. That should be self-explanatory.
Richard Webb
INdeed, but you probably wouldn't be surprised at the number of times I've seen, both in this group, and in
alt.audio.pro.live-sound folks asking which scale to use on
their rat shack meter they bought to see how loud the band
is at foh position <g>. From the nature of the questions
the op usually asks I figured Scott was a good one to
expound on the subject a bit, because he goes into enough
detail that the neophyte can understand it. YOu and I both
have the habit of assuming folks understand what they really don't.
NEither one of us were cut out to be teachers methinks <g>.
Mike Rivers
> WOuld you care to expound on when one might use a
> weighted and when C weighted, since that would be his next
> question, and would probably be beneficial.
A weighting, because it approximates the sensitivity of the
human ear, is what's used for determining a single number
for sound level that's typically used to determine a harmful
sound environment. C weighting, though not flat, is the
closest to flat on the low end, with a 10 dB hump around
about 3 kHz, tapering down linearly above that so it's back
to 0 dB around 15 kHz.
C weighting is most useful for determining the maximum SPL
in PA and theater systems, and is sometimes used in long
term monitoring. People trying to escape from the Music
Police like A weighting because it attenuates of a lot of
the low end that gets you dancing.
--
"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
Scott Dorsey
>On Tue 2011-Jul-05 13:38, William Sommerwerck writes:
>
>>> Would you care to expound on when one might use a
>>> weighted and when C weighted, since that would be his next
>>> question, and would probably be beneficial? In fact, if we
>>> were still updating the FAQ that might be a good addition.
>
>> As I and others said, the A weighting mimics the response of the ear
>> at low levels. That should be self-explanatory.
>
>INdeed, but you probably wouldn't be surprised at the number of times I've seen, both in this group, and in
>alt.audio.pro.live-sound folks asking which scale to use on
>their rat shack meter they bought to see how loud the band
>is at foh position <g>. From the nature of the questions
>the op usually asks I figured Scott was a good one to
>expound on the subject a bit, because he goes into enough
>detail that the neophyte can understand it. YOu and I both
>have the habit of assuming folks understand what they really don't.
>NEither one of us were cut out to be teachers methinks <g>.
The problem is, I used to know the answer, but I am less and less sure.
The A weighting is supposed to measure the audibility of a sound, but
in fact that's not really as useful as you think. It is a very poor
measure of the annoyance level of a sound or of the hazard level of it.
The C weighting is supposed to be as close to flat as was possible with
a particular kind of ceramic microphone that no longer exists. Modern
meters (which are capable of much better low frequency response) are
basically handicapped by making them match up to the C weighting.
It's probably a better measure of the annoyance level of a sound.
BUT, the noise ordinances in most cities are based on A-weighted
measurements. Maybe this is because of lobbying from nightclub owners
whose clubs leak a lot of low end. Maybe it's because the A-weighting
is intended to measure audibility. But it makes those ordinances useless
against low frequency sources (like stamping mills and nightclubs).
However, determining legality under those ordinances is probably the only
good use for A-weighting these days.
Really, neither one of them do what people want them to do today.
hank alrich
> On Tue 2011-Jul-05 13:38, William Sommerwerck writes:
>
> >> Would you care to expound on when one might use a
> >> weighted and when C weighted, since that would be his next
> >> question, and would probably be beneficial? In fact, if we
> >> were still updating the FAQ that might be a good addition.
>
> > As I and others said, the A weighting mimics the response of the ear
> > at low levels. That should be self-explanatory.
>
> INdeed, but you probably wouldn't be surprised at the number of times I've
> seen, both in this group, and in alt.audio.pro.live-sound folks asking
> which scale to use on their rat shack meter they bought to see how loud
> the band is at foh position <g>. From the nature of the questions the op
> usually asks I figured Scott was a good one to expound on the subject a
> bit, because he goes into enough detail that the neophyte can understand
> it. YOu and I both have the habit of assuming folks understand what they
> really don't. NEither one of us were cut out to be teachers methinks <g>.
Back in the 1970's if you opened one of those Radio Shack SPL meters
you'd see the H-P logo on the circuit board! Mine from that era got
swiped.
--
shut up and play your guitar * http://hankalrich.com/
http://www.youtube.com/watch?v=NpqXcV9DYAc
http://www.sonicbids.com/HankandShaidri
FL
have their basis in the so-called "Equal Loudness Curves" first
developed by Fletcher and Munson, and later refined by Robinson and
Dadson. Research was conducted to determine how loud sine waves of
various frequencies would have to be, to sound as loud as a 1 kHz.
sine tone, to an average listener with normal hearing (whatever that
means...).
Given that the threshold of audibility is defined as 2 * 10^-5
pascals, and that is for a 1 kHz. sine tone, a person with normal
hearing would find that a sine tone of 100 Hz. would have to be
boosted roughly 20 dB to be audible. The ear is more sensitive to
sounds in the range around 3 - 5 kHz. so those sounds could actually
be a little quieter to be "just audible" - by some 6 dB.
The equal loudness curve is thus determined to be a range of differing
sound pressures plotted against frequency, against the reference 1
kHz. tone. This collection of levels can be expressed as a curve and
assigned the value of 0 phons. Most people seem to agree that raising
the level by 10 dB results in a sound that is "twice as loud" - a very
subjective expression, but let's go with it.
Raise the 1 kHz. tone 10 dB, and you will find you need to express
equivalent loudness with a whole new range of sound pressure vs.
frequency values, which would then become the 10 phon level. And so
on, and so on in 10 phon increments until you reach the upper limit of
audibility, or the threshold of pain, at roughly 120 phons.
For example, suppose you have a 1 kHz. sine tone at 100 dB SPL. In
order for a person with normal hearing to experience a 100 Hz. tone as
being of the same loudness, the 100 Hz. tone would have to be 3 dB
higher. Make everything 50 dB lower, and the 100 Hz. tone would have
to be 8 dB louder than the 1 kHz. tone to be perceived as being of
equal loudness. - a difference of 5 dB from the condition at 100
phons.
When sound level meters were started to be produced, it was determined
that they would be put to best use if they would express a single
number for any given sound pressure condition, yet we've seen that
human hearing is particularly non-linear so that a frequency weighting
at one level of loudness would not be applicable at a louder or softer
level. As a means of simplifying both construction and operation,
three levels were selected: "A" weighting corresponds to the equal
loudness contour at 40 phons; "B" weighting, to 70 phons; and "C"
weighting to 100 phons and higher.
The "C" scale is relatively flat, but still slightly deficient at
frequencies below 100 Hz. and over 3 kHz. The "B" scale is similar to
the "C" scale at high frequencies, but is less sensitive to
frequencies below 300 hz. and is down roughly 6 dB at 100 Hz. and 10
dB at 50 Hz., relative to the "C" weighting. The ever popular "A"
weighting is roughly 3 dB MORE sensitive to frequencies above 1.2
kHz., compared to the "B" and "C" weightings, and much less sensitive
to frequencies below 1 kHz. such that it is 10 dB less sensitive to
"C" at 200 Hz., 20 dB less sensitive to "C" at 100 Hz. and 30 dB less
sensitive to "C" at 50 Hz.
The issue, as I see it, is the determination of just what constitutes
"normal" hearing, in our ever more acoustically challenged society.
The divergence between what a human experiences and what is measurable
by a device is further compounded when you take into account the
various forms that "listening" has taken: in a purely acoustic
environment, in enclosed spaces with or without amplification with
varying numbers of playback channels, via the telephone or cel phone,
or on headphones.
In short, the "A" weighting is probably the most commonly used - and
misused for higher levels than that for which it was designed. As has
been pointed out, this benefits those wishing to pump out lots of low
frequency sound, yet remain within the limits of local noise bylaws.
For a more exacting application, such as calibrating a monitoring
environment in a studio control room, particularly if you are
incorporating Bob Katz's "K" metering system, the "C" weighting curve
is the one to use to determine a sound pressure level of 83 dB (per
channel, slow speed SPL meter response) at the listening position.
Bob Katz has much more to say on this subject on his website,
particularly at
http://www.digido.com/level-practices-part-2-includes-the-k-system.html
He also has made available a pink noise WAV file, that can be used to
calibrate monitoring for the K system.
0jun...@bellsouth.net
On 2011-07-06 klu...@panix.com(ScottDorsey) said:
>>INdeed, but you probably wouldn't be surprised at the number of
>times I've seen, both in this group, and in
>>alt.audio.pro.live-sound folks asking which scale to use on
>>their rat shack meter they bought to see how loud the band
>>is at foh position <g>. From the nature of the questions
>>the op usually asks I figured Scott was a good one to
>>expound on the subject a bit, because he goes into enough
>>detail that the neophyte can understand it. YOu and I both
>>have the habit of assuming folks understand what they really don't.
>>NEither one of us were cut out to be teachers methinks <g>.
>The problem is, I used to know the answer, but I am less and less
>sure.
ordinances were based on A weighting, cbut was always under
the assumption that C weighting was more of a true
representation.
>The A weighting is supposed to measure the audibility of a sound,
>but in fact that's not really as useful as you think. It is a very
>poor measure of the annoyance level of a sound or of the hazard
>level of it.
This part I understood, but your explanation below cleared
up a couple things in my brain.
>The C weighting is supposed to be as close to flat as was possible
>with a particular kind of ceramic microphone that no longer exists.
>Modern meters (which are capable of much better low frequency
>response) are basically handicapped by making them match up to the
>C weighting. It's probably a better measure of the annoyance level
>of a sound.
>BUT, the noise ordinances in most cities are based on A-weighted
>measurements. Maybe this is because of lobbying from nightclub
>owners whose clubs leak a lot of low end. Maybe it's because the
>A-weighting is intended to measure audibility. But it makes those
>ordinances useless against low frequency sources (like stamping
>mills and nightclubs). However, determining legality under those
>ordinances is probably the only good use for A-weighting these days.
>Really, neither one of them do what people want them to do today.
There again, a good explanation, worthy of being added to
the newsgroup faq <hint for faq maintainers if any are out
there>.
Richard webb,
replace anything before at with elspider
ON site audio in the southland: see www.gatasound.com
William Sommerwerck
as a "loudness meter" that gives an accurate representation of the
subjective loudness of sounds. I don't remember the details of how it works,
but it is rather more complex than a simple filter+meter.
http://en.wikipedia.org/wiki/Loudness_monitoring
John Williamson
As it's by the same people that produce the Optimod, it might be worth
checking the manual or just playing to see how it works. It's a free
download, so all you lose is some time.
--
Tciao for Now!
John.
Arny Krueger
"FL" <lockwoo...@gmail.com> wrote in message
news:4c093a58-653e-488c...@n28g2000vbs.googlegroups.com...
<snip lots of good stuff>
> In short, the "A" weighting is probably the most commonly used - and
> misused for higher levels than that for which it was designed. As has
> been pointed out, this benefits those wishing to pump out lots of low
> frequency sound, yet remain within the limits of local noise bylaws.
The use of "A" weighting at higher levels is sometimes abuse of a good
standard, as you say.
The low frequency part of 'A' weighting makes some sense when we are
estimating accpetable SPLs as related to ear damage because the ear is far
more tolerant of high SPL's at low frequencies.
Scott Dorsey
It is an excellent piece of software with good documentation too.
But if you want to know more than you ever thought there was to know
about metering, buy "Audio Metering" by Eddy Brixen. It is really a
surprisingly interesting book.
Mike Rivers
> But if you want to know more than you ever thought there was to know
> about metering, buy "Audio Metering" by Eddy Brixen. It is really a
> surprisingly interesting book.
I just wrote a review of that book. I really wanted to like
it, and it's indeed packed with useful information, but I
just couldn't recommend it for a beginner because it goes
all over the place. It's good for looking up something to
learn more about it, but hard to just pick it up and learn
about audio metering.
ChrisCoaster
> "ChrisCoaster" <ckozi...@snet.net> wrote in message
>
> news:15a95a88-3336-400b...@x41g2000yqd.googlegroups.com...
>
> > I know that A-weighting approximates human sensitivity and C-weighting
> > is "flat".
>
> > But which one is "pink" and which is "white"??
>
> > For once and for all set me straight on the terminology.
>
> > -CC
>
>
> None of these are white, though they are all approximately white at the
> left, low end of the graph. True white noise is a mathematical curiosity. It
> is defined as equal power graphed against linear frequency. Since frequency
> goes up forever, a true white noise signal would have infinite power. So all
> the graphs convert to something other than linear as the frequency goes up.
> As soon as the curve goes horizontal against a log scale of frequency, it is
> not mathematically white.
>
> Now look at "C". "C" is not linear per frequency, but it is linear per
> octave: same power in each octave, between 100 and 3000 Hz. This is not
> white noise, but someone decided it makes good practical sense to define
> such a curve as flat.
>
> Bob Morein
> (310) 237-6511
What I want is very simple:
Pink = A = flat curve
White = C = downward slope to the left, peak center right between
3>5kHz, sloping to the right above 8kHz.(kind of like a lopsided Swiss
alp).
Simple.
-CC
Trevor
>What I want is very simple:
>White = C ...
> Simple.
Simple, but wrong!
Trevor.
ChrisCoaster
> "ChrisCoaster" <ckozi...@snet.net> wrote in message
>
> news:08ead2bc-355a-4388-935b-
>
> >What I want is very simple:
> >Pink = A ...
> >White = C ...
> > Simple.
>
> Simple, but wrong!
>
> Trevor.
Then please reaarange my equivalents in the correct order, and I will
print it out 40point font and hang it on my wall. :D
-CC
William Sommerwerck
There are no such equivalents, any more than saying an apple is equivalent
to an airplane, or an orange to a tire iron. The weighting curve has no
connection of any sort with the spectral distribution of the noise signal.
Scott Dorsey
>
>Pink =3D A =3D flat curve
>
>White =3D C =3D downward slope to the left, peak center right between
>3>5kHz, sloping to the right above 8kHz.(kind of like a lopsided Swiss
>alp).
>
>Simple.
Not really, because it depends on how you plot the curves.
ChrisCoaster
Oh no Scott - those were just examples of how I wanted the
equivalencies laid out for me. :)
The color - the weight - and whether the curve was flat or represented
human hearing.
-CC
Trevor
On Jul 8, 9:52 am, klu...@panix.com (Scott Dorsey) wrote:
> ChrisCoaster <ckozi...@snet.net> wrote:
> >Pink =3D A =3D flat curve
> >White =3D C =3D downward slope to the left, peak center right between
> >3>5kHz, sloping to the right above 8kHz.(kind of like a lopsided Swiss
> >alp).
>
> >Simple.
>
> Not really, because it depends on how you plot the curves.
}Oh no Scott - those were just examples of how I wanted the
}equivalencies laid out for me. :)
}The color - the weight - and whether the curve was flat or represented
}human hearing.
There are no such simple equivalencies (as you have been told already)
You have also been given definitions of white and pink noise spectral
response, plus A,B and C weighting curves, not to mention the
Fletcher-Munson curves.
I suggest you make your own charts however you think is appropriate for your
needs. Of course you might need to know those first!
Trevor.
Scott Dorsey
>Oh no Scott - those were just examples of how I wanted the
>equivalencies laid out for me. :)
>The color - the weight - and whether the curve was flat or represented
>human hearing.
It's not like that, though. The curves look different depending on how
you plot them. You need to listen to them and hear how they sound.
They have nothing to do with representations of human hearing. They
are mathematical distributions.
They have no connection whatsoever with weighting curves for measurements.
Scott Dorsey
>On 7/7/2011 10:02 AM, Scott Dorsey wrote:
>
>> But if you want to know more than you ever thought there was to know
>> about metering, buy "Audio Metering" by Eddy Brixen. It is really a
>> surprisingly interesting book.
>
>I just wrote a review of that book. I really wanted to like
>it, and it's indeed packed with useful information, but I
>just couldn't recommend it for a beginner because it goes
>all over the place. It's good for looking up something to
>learn more about it, but hard to just pick it up and learn
>about audio metering.
Yeah, that's true. It has everything in one place, but it has
an awful lot all in one place.
So where is your review appearing?
Mike Rivers
>> On 7/7/2011 10:02 AM, Scott Dorsey wrote:
>>> But if you want to know more than you ever thought there was to know
>>> about metering, buy "Audio Metering" by Eddy Brixen.
> So where is your review appearing?
http://mikeriversaudio.files.wordpress.com/2011/06/audiometeringbookreview.pdf
alex
> I know that A-weighting approximates human sensitivity and C-weighting
> is "flat".
>
> But which one is "pink" and which is "white"??
>
> For once and for all set me straight on the terminology.
>
> -CC
none of them. No direct relation.
A "weighting" curve is a specific kind of *equalization* *curve*, while
white and pink noise are randomly generated *signals* with a specific
equal power content per frequency or octave.
"c" weighting is not really a FLAT curve but contains rolloffs for highs
and lows well inside the hearing range.
Weighting curves are used to "correct" the measurement microphone output
to better fit some particular measurement purpuse.
"a" weighting is derived by the family of human ear sensibility curves
(although the concept of human sensibility is more complicated and
depends largely on the nature of the sound itself), and help the
measurement to best fit the way humans hear sound, which is not flat...
take a look to the fletcher-munson curves.
alex