Waveform horror! Is a 44.1 kHz sampling rate sufficient?

[Michael]

I used Audacity to generate a 19 kHz tone, planning on burning it to a
CD and playing it in my car stereo for... um... well, let's not get
into that. I then zoomed in on the waveform to take a peek, prior to
burning the waveform. What horror!

http://mrdarrett.googlepages.com/distort.jpg

The waveform is severely distorted. I could see where the points
mathematically
would be correct, but given the sparse sampling points, the speaker
would not be instructed to swing rail-to-rail at, say, between 15.0000
and 15.0005 seconds.

A FFT gives a spread of frequencies centered around the 19 kHz, but,
yuck!

I tried various other frequencies: 9 kHz gives a pretty distorted
waveform as well.

What are the implications as far as accurate sound reproduction at a
44.1 kHz
sampling frequency, as used by CDs?

Michael

 

[Jon Slaughter]

"Michael" <mrdarrett@gmail.com> wrote in message
news:609e9c8b-ca5e-4718-bf0e-f3730c6536f1@q40g2000prh.googlegroups.com...

I used Audacity to generate a 19 kHz tone, planning on burning it to a
CD and playing it in my car stereo for... um... well, let's not get
into that. I then zoomed in on the waveform to take a peek, prior to
burning the waveform. What horror!

http://mrdarrett.googlepages.com/distort.jpg

The waveform is severely distorted. I could see where the points
mathematically
would be correct, but given the sparse sampling points, the speaker
would not be instructed to swing rail-to-rail at, say, between 15.0000
and 15.0005 seconds.

A FFT gives a spread of frequencies centered around the 19 kHz, but,
yuck!

I tried various other frequencies: 9 kHz gives a pretty distorted
waveform as well.

What are the implications as far as accurate sound reproduction at a
44.1 kHz
sampling frequency, as used by CDs?

Michael

The points you are seeing that look distorted is because of the sparse
sampling and it is not a divisor of 44.1khz. The spline representation is
simply not going to show a perfect sine wave. If you were to try to fit a
sine wave to that data you would find it would fit really well. But because
the peaks are not sampled in some cases it looks like something entirely
different(related to aliasing).

The data though is still a sine wave and will sound like one. The fourier
transform(the spectrum) of any pure signal is not pure and will always be a
"mash" of frequencies around the center frequency(this is because of
numerical errors in the calculations).

I wouldn't worry about it. If you get a better program that uses a better
interpolation method then it won't look bad.

 

[Jamie]

Michael wrote:

I used Audacity to generate a 19 kHz tone, planning on burning it to a
CD and playing it in my car stereo for... um... well, let's not get
into that. I then zoomed in on the waveform to take a peek, prior to
burning the waveform. What horror!

http://mrdarrett.googlepages.com/distort.jpg

The waveform is severely distorted. I could see where the points
mathematically
would be correct, but given the sparse sampling points, the speaker
would not be instructed to swing rail-to-rail at, say, between 15.0000
and 15.0005 seconds.

A FFT gives a spread of frequencies centered around the 19 kHz, but,
yuck!

I tried various other frequencies: 9 kHz gives a pretty distorted
waveform as well.

What are the implications as far as accurate sound reproduction at a
44.1 kHz
sampling frequency, as used by CDs?

Michael
That software isn't doing the view any favors.

Doing straight point to point lines, isn't the way the
signal is going to look on the output if you were to view that with
a real scope.

If you had a transparent overlay of a pure 19khz sine wave signal, you
would see where the points would perfectly align with these waves.

Through the hardware of your sound card, the output isn't going to
respond in that manner. Differences between sample levels dictates
output behavior as it smoothly moves +/-, depending on the sign..

 

[Phil Allison]

"Michael"

I understand that yes the software is drawing lines point-to-point,
and if you superimpose a sine wave over the dots, the dots will
align... but does the speaker know that a sine wave was the intent?


** The wave shown is generated artificially by software and does NOT
represent what any sound card, CD or DVD player ACTUALLY outputs.

Fact is - once the samples are fed through a 20 kHz low pass filter, a
perfect sine wave results.

Plus, most D to A converters create extra samples in between the ones shown
to eliminate the need for steep slope filtering.

What you have to remember is that a steady sine wave only has 3 parameters:
frequency, amplitude & phase.

The smooth shape of the wave is automatic if the harmonics are filtered out
or are simply at inaudible frequencies as they ARE in your naive example.



..... Phil

 

[John Larkin]

On Sun, 20 Sep 2009 14:54:26 -0700 (PDT), Michael
<mrdarrett@gmail.com> wrote:

I used Audacity to generate a 19 kHz tone, planning on burning it to a
CD and playing it in my car stereo for... um... well, let's not get
into that. I then zoomed in on the waveform to take a peek, prior to
burning the waveform. What horror!

http://mrdarrett.googlepages.com/distort.jpg

The waveform is severely distorted. I could see where the points
mathematically
would be correct, but given the sparse sampling points, the speaker
would not be instructed to swing rail-to-rail at, say, between 15.0000
and 15.0005 seconds.

A FFT gives a spread of frequencies centered around the 19 kHz, but,
yuck!

I tried various other frequencies: 9 kHz gives a pretty distorted
waveform as well.

What are the implications as far as accurate sound reproduction at a
44.1 kHz
sampling frequency, as used by CDs?

Michael

An audio player converts the digital samples to voltages using a DAC,
and the DAC is followed by a lowpass filter. The filter effectively -
almost intelligently - interpolates the points into a smooth curve.

Shannon's Sampling Theorem says that if you lowpass filter a signal to
below F/2, sample it at frequency F, and run the resulting samples
through another filter of below F/2, the resulting output signal is a
perfect copy of the input for all signals that make it through the
filters. It looks like hell halfway through, but it works.

Burn your CD and play it through a stereo system and look at the
waveform on a scope. A 19 KHz sine wave will look fine.

John

 

[Michael]

On Sep 20, 3:18 pm, "Jon Slaughter" <Jon_Slaugh...@Hotmail.com> wrote:

"Michael" <mrdarr...@gmail.com> wrote in message

news:609e9c8b-ca5e-4718-bf0e-f3730c6536f1@q40g2000prh.googlegroups.com...



I used Audacity to generate a 19 kHz tone, planning on burning it to a
CD and playing it in my car stereo for... um... well, let's not get
into that.  I then zoomed in on the waveform to take a peek, prior to
burning the waveform.  What horror!

http://mrdarrett.googlepages.com/distort.jpg

The waveform is severely distorted.  I could see where the points
mathematically
would be correct, but given the sparse sampling points, the speaker
would not be instructed to swing rail-to-rail at, say, between 15.0000
and 15.0005 seconds.

A FFT gives a spread of frequencies centered around the 19 kHz, but,
yuck!

I tried various other frequencies:  9 kHz gives a pretty distorted
waveform as well.

What are the implications as far as accurate sound reproduction at a
44.1 kHz
sampling frequency, as used by CDs?

Michael

The points you are seeing that look distorted is because of the sparse
sampling and it is not a divisor of 44.1khz. The spline representation is
simply not going to show a perfect sine wave. If you were to try to fit a
sine wave to that data you would find it would fit really well. But because
the peaks are not sampled in some cases it looks like something entirely
different(related to aliasing).

The data though is still a sine wave and will sound like one. The fourier
transform(the spectrum) of any pure signal is not pure and will always be a
"mash" of frequencies around the center frequency(this is because of
numerical errors in the calculations).

I wouldn't worry about it. If you get a better program that uses a better
interpolation method then it won't look bad.

I understand that yes the software is drawing lines point-to-point,
and if you superimpose a sine wave over the dots, the dots will
align... but does the speaker know that a sine wave was the intent?

For example, in the time between 15.00000 and 15.00003 seconds as
shown in the pic, which would be more likely:

1. The speaker will only move a little bit (taking it easy), seeing
only a slight change in position was requested; or

2. The speaker cone's momentum will force it to take a sine wave
shape?

Thanks,

Michael

 

[Jasen Betts]

On 2009-09-20, Michael <mrdarrett@gmail.com> wrote:

I used Audacity to generate a 19 kHz tone, planning on burning it to a
CD and playing it in my car stereo for... um... well, let's not get
into that.

(hmm, fm sterio subcarrier...)

I then zoomed in on the waveform to take a peek, prior to
burning the waveform. What horror!

http://mrdarrett.googlepages.com/distort.jpg

hmm yes that does look ugly,

The waveform is severely distorted. I could see where the points
mathematically
would be correct, but given the sparse sampling points, the speaker
would not be instructed to swing rail-to-rail at, say, between 15.0000
and 15.0005 seconds.

A FFT gives a spread of frequencies centered around the 19 kHz, but,
yuck!

It's not as bad as it looks the Nyquist-shannon sampling thoerem says
in essence that signals with bandwidth less than half the sampling
frequency can be represented without loss by a set of points like
those shown.

(for the spectrum use the "gausian(a=4.5) window" function and select
a large size)

It would be nice if audacity could do a better curve-fit when it
interpolates between points instead of the naiive straight-line it's
doing. (I have submitted a request for this feature)

I tried various other frequencies: 9 kHz gives a pretty distorted
waveform as well.

it only looks bad, it won't sound bad (I mean it won't sound significantly
worse than a perfect 9 or 19khz sine wave).

 

[Phil Allison]

"Jasen Betts" <jasen@xnet.co.nz>

= ASD fucked sheep shagging FAGGOT



" thoerem " " gausian " " naiive " ????????


** Jesus H. Bloody Christ ....




...... Phil

 

[Rich Webb]

On Mon, 21 Sep 2009 22:15:21 +1000, "Phil Allison" <phil_a@tpg.com.au>
wrote:

"Jasen Betts" <jasen@xnet.co.nz

= ASD fucked sheep shagging FAGGOT

Wut's an ASD, guv? ^_^

--
Rich Webb Norfolk, VA

 

[Phil Allison]

"Rich Webb"
"Phil Allison"

"Jasen Betts" <jasen@xnet.co.nz

= ASD fucked sheep shagging FAGGOT

Wut's an ASD, guv?

** Never heard of Google ??

Typical of the terminally ASD fucked to have no clue as to terminology of
their own condition.





...... Phil

 

[Jon Slaughter]

"Michael" <mrdarrett@gmail.com> wrote in message
news:3e5d0f46-649a-4652-b123-132346260606@y10g2000prg.googlegroups.com...
On Sep 20, 3:18 pm, "Jon Slaughter" <Jon_Slaugh...@Hotmail.com> wrote:

"Michael" <mrdarr...@gmail.com> wrote in message

news:609e9c8b-ca5e-4718-bf0e-f3730c6536f1@q40g2000prh.googlegroups.com...



I used Audacity to generate a 19 kHz tone, planning on burning it to a
CD and playing it in my car stereo for... um... well, let's not get
into that. I then zoomed in on the waveform to take a peek, prior to
burning the waveform. What horror!

http://mrdarrett.googlepages.com/distort.jpg

The waveform is severely distorted. I could see where the points
mathematically
would be correct, but given the sparse sampling points, the speaker
would not be instructed to swing rail-to-rail at, say, between 15.0000
and 15.0005 seconds.

A FFT gives a spread of frequencies centered around the 19 kHz, but,
yuck!

I tried various other frequencies: 9 kHz gives a pretty distorted
waveform as well.

What are the implications as far as accurate sound reproduction at a
44.1 kHz
sampling frequency, as used by CDs?

Michael

The points you are seeing that look distorted is because of the sparse
sampling and it is not a divisor of 44.1khz. The spline representation is
simply not going to show a perfect sine wave. If you were to try to fit a
sine wave to that data you would find it would fit really well. But
because
the peaks are not sampled in some cases it looks like something entirely
different(related to aliasing).

The data though is still a sine wave and will sound like one. The fourier
transform(the spectrum) of any pure signal is not pure and will always be
a
"mash" of frequencies around the center frequency(this is because of
numerical errors in the calculations).

I wouldn't worry about it. If you get a better program that uses a better
interpolation method then it won't look bad.

I understand that yes the software is drawing lines point-to-point,
and if you superimpose a sine wave over the dots, the dots will
align... but does the speaker know that a sine wave was the intent?

For example, in the time between 15.00000 and 15.00003 seconds as
shown in the pic, which would be more likely:

1. The speaker will only move a little bit (taking it easy), seeing
only a slight change in position was requested; or

2. The speaker cone's momentum will force it to take a sine wave
shape?
--

Yes, for the most part, if the speaker is designed to hangle that frequency
and all the drive circuitry can drive the cone exactly. Speakers have
inertia which prevents them from oscillating at arbitrarily high
frequencies. Hence the reason why you have subwoofers and tweeters.

There are "horns" that are much better for such things.

---

The speaker design is basically a forced oscillator. As you probably know,
the mechanical system is like a spring with a mass attached. The
mathematical analysis of this is a differential equation. The system is
damped due to inertia and friction. The driving force is the coil attached
which is driven by the audio amplifier. For such a system there will be a
transient response and then the system tends to follow the forcing
function(at least if it's somewhat regular).

So It's a property of the system itself to follow the audio. If the
transients are quickly removed and the mechanical properties of inertia and
friction are balanced with the "spring" then the speaker will follow the
audio rather well.

So what this is saying is that the "momentum" will make it follow a sine
wave. It's really the spring effect designed in the speaker and the inertia
of the cone. Because it is a spring, and if you have any experience with
mechanical oscillators, you tend to know that they oscillate "sinusoidally".

Now, the real question is how perfect is it? If it does not oscillate
perfectly then you have "harmonics". If the system is sufficiently dampened
then the harmonics can be reduced significanly. What is left is a pure sine
wave.

This isn't intentionally done in a speaker though because then you would not
hear the difference between a sine wave and a square wave. Essentially a
"filtering" occurs which is not then reproducing the audio intentionally.

BUT, note that for a square wave we have to have a much higher sampling rate
to reproduce it. So at your 19khz, for the speaker to produce a good square
wave, we need a sampling and driving system that runs much higher(usually at
least a factor of 10). For sin waves it only needs to be a factor of 2. So
you can rest assured that the output will not be a square wave or close. It
might not be a perfect sin wave but this is the "natural" response of the
system both mathematically and mechanically.

But ultimately it depends on the design. If you have a bad design it could
create harmonics(a non-linear system). Such design might be where the
speaker somehow has friction in it that behaves oddly. Maybe a tear in the
cone that catches against the base in some odd and chaotic way.


My guess, as I'm no expert in such matters, is that if the speaker is
capable of such high frequencies then it would work well. You most likely
won't get the sound out of a subwoofer because of the large amount of
inertia. The speaker simply can't move fast enough and could end up actually
oscillating at a lower frequency or not oscillate at all. If you find the
right way to reproduce the sound you can get it very accurately.
Electrically you can get extremely precise sound wave reproduction.

If your goal is to reproduce such a pure sound wave then you might try to
create a tunable electrical oscillator. This is quite easy to do, although
much more complex than playing a cd, and will be very accurate. You can also
get loads of power for it to drive any system you want and could adjust the
frequency to your liking.

 

[Tim Wescott]

On Mon, 21 Sep 2009 13:24:13 +1000, Phil Allison wrote:

"Michael"

I understand that yes the software is drawing lines point-to-point, and
if you superimpose a sine wave over the dots, the dots will align... but
does the speaker know that a sine wave was the intent?


** The wave shown is generated artificially by software and does NOT
represent what any sound card, CD or DVD player ACTUALLY outputs.

Fact is - once the samples are fed through a 20 kHz low pass filter, a
perfect sine wave results.

Plus, most D to A converters create extra samples in between the ones
shown to eliminate the need for steep slope filtering.

What you have to remember is that a steady sine wave only has 3
parameters: frequency, amplitude & phase.

The smooth shape of the wave is automatic if the harmonics are filtered
out or are simply at inaudible frequencies as they ARE in your naive
example.

If anyone but you wrote that you'd be off on a foam at the mouth,
vituperative foul-mouthed rant.

Some of what you say is correct on it's face, but highly misleading. The
rest is just plane wrong.

If you ran that signal through a _perfect_ low pass filter with an
absolute cutoff at 20kHz, the result would, indeed, be a perfect sine
wave. But such filters don't exist. A 19kHz signal that's sampled at
44.1kHz will have content at 19kHz, 25.1kHz, and every harmonic of
44.1kHz +/- 19kHz. A _real_ low pass reconstruction filter will
attenuate the 19kHz signal to some extent, and the 25.1kHz signal to a
much greater extent -- but the results won't be 'perfect'.

Perhaps most _audio_ D/A converters will do oversampling, but those extra
samples still need to come from somewhere, that 'where' is a low pass
filter, and all my comments about low pass filters apply. Only a D/A
converter that is specifically designed for audio will have a
reconstruction filter that can be expected to do a good job with this
signal.

You are correct that a steady sine wave can be characterized by just
three parameters, at least, but you imply that there's nothing more to
this discussion, which is incorrect.

And yes, the smooth shape of the wave is automatic if the harmonics are
filtered out, and yes _most_ people aren't going to be able to hear the
25.1kHz signal, at least by itself, and yes, if this is getting played
through a car stereo and listened to by humans then you're probably
mostly right.

But if the OP tries to do something other than play this through a piece
of audio equipment for human consumption then he may be quite
disappointed.

--
www.wescottdesign.com

 

[Tim Wescott]

On Sun, 20 Sep 2009 14:54:26 -0700, Michael wrote:

I used Audacity to generate a 19 kHz tone, planning on burning it to a
CD and playing it in my car stereo for... um... well, let's not get into
that. I then zoomed in on the waveform to take a peek, prior to burning
the waveform. What horror!

http://mrdarrett.googlepages.com/distort.jpg

The waveform is severely distorted. I could see where the points
mathematically
would be correct, but given the sparse sampling points, the speaker
would not be instructed to swing rail-to-rail at, say, between 15.0000
and 15.0005 seconds.

A FFT gives a spread of frequencies centered around the 19 kHz, but,
yuck!

I tried various other frequencies: 9 kHz gives a pretty distorted
waveform as well.

What are the implications as far as accurate sound reproduction at a
44.1 kHz
sampling frequency, as used by CDs?

Michael

Part of the theory of CDs is that sound reproduction doesn't need to be
accurate at that high of a frequency range, because the human ear doesn't
hear that well up there.

When this signal is run through a reconstruction filter it'll end up
being more or less nicely sinusoidal. IF you're just playing it through
your car stereo so that folks can hear a 19kHz tone then you're probably
OK, assuming the CD's compression doesn't mangle it. If you want
accurate reproduction at 19kHz for the purposes of some machine 'hearing'
it and doing the right thing, you may be disappointed.

This may help your understanding of the sampling and reconstruction
phenomena: http://www.wescottdesign.com/articles/Sampling/
sampling.html. It's mostly about debunking erroneous conclusions of the
Nyquist-Shannon sampling theorem, but it's got the background in there
for you to understand what's going on with your signal.

--
www.wescottdesign.com

 

[Tim Wescott]

On Mon, 21 Sep 2009 09:41:57 -0700, Michael wrote:

On Sep 21, 9:24 am, Tim Wescott <t...@seemywebsite.com> wrote:
On Sun, 20 Sep 2009 14:54:26 -0700, Michael wrote:
I used Audacity to generate a 19 kHz tone, planning on burning it to
a CD and playing it in my car stereo for... um... well, let's not get
into that.  I then zoomed in on the waveform to take a peek, prior to
burning the waveform.  What horror!

http://mrdarrett.googlepages.com/distort.jpg

The waveform is severely distorted.  I could see where the points
mathematically
would be correct, but given the sparse sampling points, the speaker
would not be instructed to swing rail-to-rail at, say, between
15.0000 and 15.0005 seconds.

A FFT gives a spread of frequencies centered around the 19 kHz, but,
yuck!

I tried various other frequencies:  9 kHz gives a pretty distorted
waveform as well.

What are the implications as far as accurate sound reproduction at a
44.1 kHz
sampling frequency, as used by CDs?

Michael

Part of the theory of CDs is that sound reproduction doesn't need to be
accurate at that high of a frequency range, because the human ear
doesn't hear that well up there.

When this signal is run through a reconstruction filter it'll end up
being more or less nicely sinusoidal.  IF you're just playing it
through your car stereo so that folks can hear a 19kHz tone then you're
probably OK, assuming the CD's compression doesn't mangle it.  If you
want accurate reproduction at 19kHz for the purposes of some machine
'hearing' it and doing the right thing, you may be disappointed.


I chose 19 kHz so that dogs can hear it, but humans above about 20 years
of age cannot. I also chose the frequency to be below 20 kHz, which
IIRC is the limit of most speakers.

Michael

That 20kHz "limit" is pretty soft as far as actual performance goes --
the manufacturer assigns it a hard number because that's what a kid who
can still hear 19kHz can understand.

Imagine a brick wall. Imagine walking into a brick wall. It's pretty
obvious where the line is between "you can walk here" and "you can't walk
here".

Now imagine the bank of a flat river, with dry grassland merging smoothly
into water meadows and the soil level of the water meadows smoothly
dropping down into the water, and the river getting gradually deeper
until it's finally 20' deep in the middle with a swift current. Where's
the line between "you can walk here" and "you can't walk here"? It
depends on what kind of shoes you have, and how determined you are, and
whether you want to get your pants muddy, doesn't it?

That 20kHz performance limit on the speakers is a _lot_ more like the
'edge' of that river than the edge of the brick wall.

Good luck, whatever you're trying. Don't deafen any kids...

--
www.wescottdesign.com

 

[John Fields]

On Mon, 21 Sep 2009 23:16:54 +1000, "Phil Allison" <phil_a@tpg.com.au>
wrote:

"Rich Webb"
"Phil Allison"
"Jasen Betts" <jasen@xnet.co.nz

= ASD fucked sheep shagging FAGGOT

Wut's an ASD, guv?


** Never heard of Google ??

Typical of the terminally ASD fucked to have no clue as to terminology of
their own condition.

---

http://www.asdk12.org/ ?

 

On Sep 21, 9:00 am, "Jon Slaughter" <Jon_Slaugh...@Hotmail.com> wrote:
<snip>

I understand that yes the software is drawing lines point-to-point,
and if you superimpose a sine wave over the dots, the dots will
align... but does the speaker know that a sine wave was the intent?

For example, in the time between 15.00000 and 15.00003 seconds as
shown in the pic, which would be more likely:

1.  The speaker will only move a little bit (taking it easy),
seeing
only a slight change in position was requested; or

2.  The speaker cone's momentum will force it to take a sine wave
shape?
--

Yes, for the most part, if the speaker is designed to hangle that
frequency
and all the drive circuitry can drive the cone exactly. Speakers
have
inertia which prevents them from oscillating at arbitrarily high
frequencies. Hence the reason why you have subwoofers and tweeters.

There are "horns" that are much better for such things.

---

The speaker design is basically a forced oscillator. As you
probably know,
the mechanical system is like a spring with a mass attached. The
mathematical analysis of this is a differential equation. The
system is
damped due to inertia and friction. The driving force is the coil
attached
which is driven by the audio amplifier.  For such a system there
will be a
transient response and then the system tends to follow the forcing
function(at least if it's somewhat regular).

So It's a property of the system itself to follow the audio. If the
transients are quickly removed and the mechanical properties of
inertia and
friction are balanced with the "spring" then the speaker will
follow the
audio rather well.

So what this is saying is that the "momentum" will make it follow a
sine
wave. It's really the spring effect designed in the speaker and the
inertia
of the cone. Because it is a spring, and if you have any experience
with
mechanical oscillators, you tend to know that they oscillate
"sinusoidally".

Now, the real question is how perfect is it? If it does not
oscillate
perfectly then you have "harmonics". If the system is sufficiently
dampened
then the harmonics can be reduced significanly. What is left is a
pure sine
wave.

This isn't intentionally done in a speaker though because then you
would not
hear the difference between a sine wave and a square wave.
Essentially a
"filtering" occurs which is not then reproducing the audio
intentionally.

BUT, note that for a square wave we have to have a much higher
sampling rate
to reproduce it. So at your 19khz, for the speaker to produce a
good square
wave, we need a sampling and driving system that runs much higher
(usually at
least a factor of 10). For sin waves it only needs to be a factor
of 2. So
you can rest assured that the output will not be a square wave or
close. It
might not be a perfect sin wave but this is the "natural" response
of the
system both mathematically and mechanically.

But ultimately it depends on the design. If you have a bad design
it could
create harmonics(a non-linear system). Such design might be where
the
speaker somehow has friction in it that behaves oddly. Maybe a tear
in the
cone that catches against the base in some odd and chaotic way.

My guess, as I'm no expert in such matters, is that if the speaker
is
capable of such high frequencies then it would work well.  You most
likely
won't get the sound out of a subwoofer because of the large amount
of
inertia. The speaker simply can't move fast enough and could end up
actually
oscillating at a lower frequency or not oscillate at all. If you
find the
right way to reproduce the sound you can get it very accurately.
Electrically you can get extremely precise sound wave reproduction.

If your goal is to reproduce such a pure sound wave then you might
try to
create a tunable electrical oscillator. This is quite easy to do,
although
much more complex than playing a cd, and will be very accurate. You
can also
get loads of power for it to drive any system you want and could
adjust the
frequency to your liking.

"Forced oscillator" ? What the heck is that? Do what John Larkin
suggested and make the CD, play it and look at the output of the
player. I know he's right because I tried that 10 years ago when I got
my first CD burner. BTW, CoolEdit / Adobe Audition shows the same
timing / magnitude of the samples but shows the filtered output which
I _know_ is correct as I've had my Tek scope connected to the computer
output to see it.

G˛

 

[Michael]

On Sep 21, 9:24 am, Tim Wescott <t...@seemywebsite.com> wrote:

On Sun, 20 Sep 2009 14:54:26 -0700, Michael wrote:
I used Audacity to generate a 19 kHz tone, planning on burning it to a
CD and playing it in my car stereo for... um... well, let's not get into
that.  I then zoomed in on the waveform to take a peek, prior to burning
the waveform.  What horror!

http://mrdarrett.googlepages.com/distort.jpg

The waveform is severely distorted.  I could see where the points
mathematically
would be correct, but given the sparse sampling points, the speaker
would not be instructed to swing rail-to-rail at, say, between 15.0000
and 15.0005 seconds.

A FFT gives a spread of frequencies centered around the 19 kHz, but,
yuck!

I tried various other frequencies:  9 kHz gives a pretty distorted
waveform as well.

What are the implications as far as accurate sound reproduction at a
44.1 kHz
sampling frequency, as used by CDs?

Michael

Part of the theory of CDs is that sound reproduction doesn't need to be
accurate at that high of a frequency range, because the human ear doesn't
hear that well up there.

When this signal is run through a reconstruction filter it'll end up
being more or less nicely sinusoidal.  IF you're just playing it through
your car stereo so that folks can hear a 19kHz tone then you're probably
OK, assuming the CD's compression doesn't mangle it.  If you want
accurate reproduction at 19kHz for the purposes of some machine 'hearing'
it and doing the right thing, you may be disappointed.

I chose 19 kHz so that dogs can hear it, but humans above about 20
years of age cannot. I also chose the frequency to be below 20 kHz,
which IIRC is the limit of most speakers.

Michael

 

[BobG]

Get a motorola super horn piezo tweeter at radioshack. They are
capacitive in nature, so they sort of self crossover and become
audible about 2khz, and I've looked at the output of a good mic in
front of one of them and they seem to chug right on up past 20k no
prob. I think they will take about 15V. All you need to perform the
experiment is a piezo, a signal generator, a couple of gator clips,
and a dog.

 

[Phil Allison]

"Tim Wescott = Insane Asshole "
Phil Allison wrote:

** The wave shown is generated artificially by software and does NOT
represent what any sound card, CD or DVD player ACTUALLY outputs.

Fact is - once the samples are fed through a 20 kHz low pass filter, a
perfect sine wave results.

Plus, most D to A converters create extra samples in between the ones
shown to eliminate the need for steep slope filtering.

What you have to remember is that a steady sine wave only has 3
parameters: frequency, amplitude & phase.

The smooth shape of the wave is automatic if the harmonics are filtered
out or are simply at inaudible frequencies as they ARE in your naive
example.

If anyone but you wrote that you'd be off on a foam at the mouth,
vituperative foul-mouthed rant.

** Fuck you - cunthead.

Some of what you say is correct on it's face, but highly misleading. The
rest is just plane wrong.

** None of it is wrong - you pedantic FUCKWIT.

If you ran that signal through a _perfect_ low pass filter with an
absolute cutoff at 20kHz, the result would, indeed, be a perfect sine
wave. But such filters don't exist.

** A 19kHz sine wave from a test disk played on a CD player looks perfect on
a scope.

That is all the OP needs to know - you fucking MORON.

If you ran that signal through a _perfect_ low pass filter with an
absolute cutoff at 20kHz, the result would, indeed, be a perfect sine
wave. But such filters don't exist. A 19kHz signal that's sampled at
44.1kHz will have content at 19kHz, 25.1kHz, and every harmonic of
44.1kHz +/- 19kHz. A _real_ low pass reconstruction filter will
attenuate the 19kHz signal to some extent, and the 25.1kHz signal to a
much greater extent -- but the results won't be 'perfect'.

** A 19kHz sine wave from a test disk played on a CD player looks perfect on
a scope.

That is all the OP needs to know - you fucking MORON.

Perhaps most _audio_ D/A converters will do oversampling, but those extra
samples still need to come from somewhere, that 'where' is a low pass
filter, and all my comments about low pass filters apply.

** Drivel.

Only a D/A
converter that is specifically designed for audio will have a
reconstruction filter that can be expected to do a good job with this
signal.

** The OP's question was ** about CD audio ** - you

STUPID FUCKING CUNTHEAD !!!!!

Quote from the OP:

" What are the implications as far as accurate sound reproduction at a 44.1
kHz sampling frequency, as used by CDs? "

You are correct that a steady sine wave can be characterized by just
three parameters, at least, but you imply that there's nothing more to
this discussion, which is incorrect.

** Nothing in my post implied any such thing.

YOU imagined it - inside you puny autism fucked brain.

And yes, the smooth shape of the wave is automatic if the harmonics are
filtered out, and yes _most_ people aren't going to be able to hear the
25.1kHz signal, at least by itself, and yes, if this is getting played
through a car stereo and listened to by humans then you're probably
mostly right.

** The post was entirely correct.

Cos the question was about the world of AUDIO.

YOU FUCKED MORON !!!!!!!!!!!!!!!!!!!

But if the OP tries to do something other than play this through a piece
of audio equipment for human consumption then he may be quite
disappointed.

** The post was entirely correct.

Cos the question was about the world of AUDIO.

YOU ASD FUCKED MORON !!!!!!!!!!!!!!!!!

Go DROP DEAD



....... Phil

 

[Phil Allison]

"Tim Wescott is a Raving Lunatic "

Part of the theory of CDs is that sound reproduction doesn't need to be
accurate at that high of a frequency range,

** Massive pile of BOLLOCKS !!!!!!!!!!!

The CD system provides perfect reproduction of all audio signals up to
20kHz.

When this signal is run through a reconstruction filter it'll end up
being more or less nicely sinusoidal. IF you're just playing it through
your car stereo so that folks can hear a 19kHz tone then you're probably
OK, assuming the CD's compression doesn't mangle it.

** Massive pile of BOLLOCKS.

Do not take ANY notice of this raving autistic fuckwit.




...... Phil