Reply
 
Thread Tools Display Modes
  #1   Report Post  
Posted to rec.audio.pro
karle karle is offline
external usenet poster
 
Posts: 4
Default Sample rate conversion question

Hi All,

I will be doing some tests, to find out if people can really hear the
difference if a song is played at different sampling rates.

Does anyone knows about a computer / Software / Hardware bundle that would
enable playing songs and changing the master clock on the fly? I would need
the DA converter to switch its clock automatically, according to the signal
it receives at the inputs and the finest equipment if possible.

Also, if you already conducted such an experiment, I'd like to know how you
did it.

Thanks,

Karle
McGill University


  #2   Report Post  
Posted to rec.audio.pro
Scott Dorsey Scott Dorsey is offline
external usenet poster
 
Posts: 16,853
Default Sample rate conversion question

karle wrote:

I will be doing some tests, to find out if people can really hear the
difference if a song is played at different sampling rates.


What kind of rates?

In general, if you make broad sweeping changes to the sampling rate, it
is usually inaudible unless there is bandlimiting going on in the process.
But, many sample rate conversion algorithms do very bad things when the
input and output rates are close to one another. You can make an AD1890
go berserk with two clocks that are only a few tens of hertz apart.

Rate conversion algorithms have changed a lot, too. Some early implementations
had some numeric precision issues that wound up creating audible artifacts.

Does anyone knows about a computer / Software / Hardware bundle that would
enable playing songs and changing the master clock on the fly? I would need
the DA converter to switch its clock automatically, according to the signal
it receives at the inputs and the finest equipment if possible.


Most outboard D/A converters will do this without any problem. Note that
most converters today are sigma-delta converters so you already have some
internal rate conversion going on inside the box.

Also, if you already conducted such an experiment, I'd like to know how you
did it.


I haven't, but Steven Harris did a bunch of listening tests when he was
with Crystal Semiconductor, and a poke at old JAES issues should turn some up.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."
  #3   Report Post  
Posted to rec.audio.pro
karle karle is offline
external usenet poster
 
Posts: 4
Default Sample rate conversion question

Thanks,

I want to know if average people can really hear the difference between,
44.1, 48, 88.2, 96, an 192 khz.

The advice I need is, should I:

1-Start with a 192khz source, downsample it and play it back at the
downsampled rates on the D to A converter
2-Start with a 192khz source, downsample it and play it back at the highest
rate on the D to A converter
3-Record the same performance at different sample rates and then play it
back at its native sample rate

Karle


"Scott Dorsey" wrote in message
...
karle wrote:

I will be doing some tests, to find out if people can really hear the
difference if a song is played at different sampling rates.


What kind of rates?

In general, if you make broad sweeping changes to the sampling rate, it
is usually inaudible unless there is bandlimiting going on in the process.
But, many sample rate conversion algorithms do very bad things when the
input and output rates are close to one another. You can make an AD1890
go berserk with two clocks that are only a few tens of hertz apart.

Rate conversion algorithms have changed a lot, too. Some early
implementations
had some numeric precision issues that wound up creating audible
artifacts.

Does anyone knows about a computer / Software / Hardware bundle that would
enable playing songs and changing the master clock on the fly? I would
need
the DA converter to switch its clock automatically, according to the
signal
it receives at the inputs and the finest equipment if possible.


Most outboard D/A converters will do this without any problem. Note that
most converters today are sigma-delta converters so you already have some
internal rate conversion going on inside the box.

Also, if you already conducted such an experiment, I'd like to know how
you
did it.


I haven't, but Steven Harris did a bunch of listening tests when he was
with Crystal Semiconductor, and a poke at old JAES issues should turn some
up.
--scott

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



  #4   Report Post  
Posted to rec.audio.pro
Laurence Payne[_2_] Laurence Payne[_2_] is offline
external usenet poster
 
Posts: 1,267
Default Sample rate conversion question

On Fri, 15 Aug 2008 12:24:51 -0400, "karle"
wrote:

1-Start with a 192khz source, downsample it and play it back at the
downsampled rates on the D to A converter
2-Start with a 192khz source, downsample it and play it back at the highest
rate on the D to A converter
3-Record the same performance at different sample rates and then play it
back at its native sample rate



Well, 2 won't be much use as it will play the sample at the wrong
pitch and speed.

InitiallyI wouldn't worry about the higher rates. Start at 48KHz,
work down from there. See at what figure people reliably start
hearing the difference.
  #5   Report Post  
Posted to rec.audio.pro
Paul Stamler Paul Stamler is offline
external usenet poster
 
Posts: 1,614
Default Sample rate conversion question

"karle" wrote in message
...

I want to know if average people can really hear the difference between,
44.1, 48, 88.2, 96, an 192 khz.

The advice I need is, should I:

1-Start with a 192khz source, downsample it and play it back at the
downsampled rates on the D to A converter
2-Start with a 192khz source, downsample it and play it back at the

highest
rate on the D to A converter
3-Record the same performance at different sample rates and then play it
back at its native sample rate


The latter. That way, you won't have the possibility of confusing the issue
with rate-conversion artifacts.

Oh, of course record everything at the same bit-depth, preferably 24 bits.
You only want one variable, and you don't want to raise the side issue of 16
bits vs. 24.

Now the tough part: how are you going to get identical performances?

Peace,
Paul




  #6   Report Post  
Posted to rec.audio.pro
Scott Dorsey Scott Dorsey is offline
external usenet poster
 
Posts: 16,853
Default Sample rate conversion question

karle wrote:

I want to know if average people can really hear the difference between,
44.1, 48, 88.2, 96, an 192 khz.


Oh, that's a totally different set of nightmares. There was a recent
JAES paper on downsampling that you might want to look at.

The thing is, you have ultrasonic audibility issues, you have sample
rate conversion artifacts. You have converter artifacts, which are
different at different rates. The confounding variables are many.

The advice I need is, should I:

1-Start with a 192khz source, downsample it and play it back at the
downsampled rates on the D to A converter


If there are artifacts with downsampling, you'll hear differences. If
there is ultrasonic perception going on, you'll hear differences.
If there are differences due to converter artifacts, you'll hear
differences.

2-Start with a 192khz source, downsample it and play it back at the highest
rate on the D to A converter


This eliminates differences due to converter artifacts changing with the
data rate but leaves the others.

3-Record the same performance at different sample rates and then play it
back at its native sample rate


This eliminates differences due to sample rate conversion artifacts but
leaves the others.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
  #7   Report Post  
Posted to rec.audio.pro
Tobiah Tobiah is offline
external usenet poster
 
Posts: 666
Default Sample rate conversion question

On Fri, 15 Aug 2008 08:54:34 -0400, karle wrote:

Hi All,

I will be doing some tests, to find out if people can really hear the
difference if a song is played at different sampling rates.

Does anyone knows about a computer / Software / Hardware bundle that would
enable playing songs and changing the master clock on the fly? I would need
the DA converter to switch its clock automatically, according to the signal
it receives at the inputs and the finest equipment if possible.


I would record your audio at the highest depth and width that you
are capable of. Make several copies of the audio, converting to
various lower rates and widths. Convert the copies back to the
high width and depth. Now you should be able to splice sections
of the different files together, perhaps at measure marks if it is to be
a song. If it is not a regular song, maybe 5 second segments
would work.

Then, publish the audio, and the divisions, and have people rate
the quality of the various sections.
** Posted from http://www.teranews.com **
  #8   Report Post  
Posted to rec.audio.pro
Mike Rivers Mike Rivers is offline
external usenet poster
 
Posts: 8,744
Default Sample rate conversion question

karle wrote:

I want to know if average people can really hear the difference between,
44.1, 48, 88.2, 96, an 192 khz.


I can answer that right now. No. The average people are very
non-discriminatory. Audio fussbudjits, however, can hear the difference
between 44.1 and 44.1 and 44.1 kHz all of the same word length.



--
If you e-mail me and it bounces, use your secret decoder ring and reach
me he
double-m-eleven-double-zero at yahoo -- I'm really Mike Rivers
)
  #9   Report Post  
Posted to rec.audio.pro
Sean Conolly Sean Conolly is offline
external usenet poster
 
Posts: 638
Default Sample rate conversion question


"Paul Stamler" wrote in message
...
"karle" wrote in message
...

I want to know if average people can really hear the difference between,
44.1, 48, 88.2, 96, an 192 khz.

The advice I need is, should I:

1-Start with a 192khz source, downsample it and play it back at the
downsampled rates on the D to A converter
2-Start with a 192khz source, downsample it and play it back at the

highest
rate on the D to A converter
3-Record the same performance at different sample rates and then play it
back at its native sample rate


The latter. That way, you won't have the possibility of confusing the
issue
with rate-conversion artifacts.

Oh, of course record everything at the same bit-depth, preferably 24 bits.
You only want one variable, and you don't want to raise the side issue of
16
bits vs. 24.

Now the tough part: how are you going to get identical performances?


How about generating the performace in the computer by using a soft synth?
Take a midi file and render it at different sample rates.

Sean


  #10   Report Post  
Posted to rec.audio.pro
Chris Hornbeck Chris Hornbeck is offline
external usenet poster
 
Posts: 1,744
Default Sample rate conversion question

On Fri, 15 Aug 2008 08:54:34 -0400, "karle"
wrote:

I will be doing some tests, to find out if people can really hear the
difference if a song is played at different sampling rates.

Does anyone knows about a computer / Software / Hardware bundle that would
enable playing songs and changing the master clock on the fly? I would need
the DA converter to switch its clock automatically, according to the signal
it receives at the inputs and the finest equipment if possible.

Also, if you already conducted such an experiment, I'd like to know how you
did it.


Until you frame the question better, you can't learn anything
interesting. You're suggesting changing a bunch of things all
together, then drawing a conclusion about one of the things
you've changed. As science, this stinks on ice.

My recommendation, FWIW, is to begin by asking what, specifically,
is different in a modern recording-and-reproduction chain between
different "sampling rates". Defining "sampling rates" is the first
hurdle. It's *not* the simple number that your post implies, (and
Scott has strongly implied the solution).

Then, building a test of the differences is comparatively easy.

All the best fortune,
Chris Hornbeck


  #11   Report Post  
Posted to rec.audio.pro
Scott Dorsey Scott Dorsey is offline
external usenet poster
 
Posts: 16,853
Default Sample rate conversion question

In article m3npk.28$5C.0@trnddc02, Mike Rivers wrote:
karle wrote:

I want to know if average people can really hear the difference between,
44.1, 48, 88.2, 96, an 192 khz.


I can answer that right now. No. The average people are very
non-discriminatory. Audio fussbudjits, however, can hear the difference
between 44.1 and 44.1 and 44.1 kHz all of the same word length.


The average person can't tell the difference between CD and a cassette
that has been dunked in orange soda.

What is most horrifying is that a friend of mine has been doing threshold
tests on college freshman as part of a psychology project, and says more
than half of them can't hear a 16 KC tone at any level.

Wear earplugs, guys. Protect your hearing, because once it is gone it does
not come back.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."
  #12   Report Post  
Posted to rec.audio.pro
karle karle is offline
external usenet poster
 
Posts: 4
Default Sample rate conversion question

HI Guys,

Thanks for the answers. I would appreciate it if people contributed to the
problem instead raising questions withou trying to answer those. I need
opinions here.
Treat this as if it were your own experiment and share your approach or
theory.

Until you frame the question better, you can't learn anything


Then please tell me how you would approach the question.

My recommendation, FWIW, is to begin by asking what, specifically,
is different in a modern recording-and-reproduction chain between
different "sampling rates". Defining "sampling rates" is the first
hurdle. It's *not* the simple number that your post implies, (and
Scott has strongly implied the solution).


It doesn't have to be complicated. It could be tedious, but not complicated.
First, we use the same brand DA or AD converter for everything. Let's assume
that I could manage to get a manufacturer to match several units to an
acceptable level. (Yes, we need to define what would be an acceptable error
between the different units)

Then, we have to decide if the following methods are acceptable:

A-downsample a sound file to different rates and then upsample back the
resultiung sound files to play everything at the same rate.
B-downsample and then play every file at its respective downsampled rate.
C-or use X number of computers (one for each sample rate to be tested),
record with the same brand of AD, feed a live performance to every computer
at the same time and those computer would each record at a different
sampling rate. And then play the files back at their native sample rate.
D-Any method someone can think of around here.

The problem with C is, what do I record, and how do I record it to make sure
that the source will be "Accepted" by the audio community as a "Benchmark"
recording?

Scott and Paul gave good ideas, I would like to have more. I need to hear
all the potential issues and questions. Then, I will be able to design a
proper experiment.

I first need to isolate the "Sampling Rate" thing . Then, investigate the
different elements in the chain.

For example, how will a mastering engineer approach a song, which would have
to be delivered at different rates (CD, SACD, whatever)
Then when a end user listens to the product, what are the different elements
in the " live source-recording-mixing-mastering-delivery medium-end user
playback unit" chain.

But first I need to isolate the sampling rate issue and decide which
processing artifacts are acceptable in such a test.

Thanks for your insights,

Karle



"Chris Hornbeck" wrote in message
...
On Fri, 15 Aug 2008 08:54:34 -0400, "karle"
wrote:

I will be doing some tests, to find out if people can really hear the
difference if a song is played at different sampling rates.

Does anyone knows about a computer / Software / Hardware bundle that would
enable playing songs and changing the master clock on the fly? I would
need
the DA converter to switch its clock automatically, according to the
signal
it receives at the inputs and the finest equipment if possible.

Also, if you already conducted such an experiment, I'd like to know how
you
did it.


Until you frame the question better, you can't learn anything
interesting. You're suggesting changing a bunch of things all
together, then drawing a conclusion about one of the things
you've changed. As science, this stinks on ice.

My recommendation, FWIW, is to begin by asking what, specifically,
is different in a modern recording-and-reproduction chain between
different "sampling rates". Defining "sampling rates" is the first
hurdle. It's *not* the simple number that your post implies, (and
Scott has strongly implied the solution).

Then, building a test of the differences is comparatively easy.

All the best fortune,
Chris Hornbeck



  #13   Report Post  
Posted to rec.audio.pro
Scott Dorsey Scott Dorsey is offline
external usenet poster
 
Posts: 16,853
Default Sample rate conversion question

karle wrote:
Thanks for the answers. I would appreciate it if people contributed to the
problem instead raising questions withou trying to answer those.


You haven't defined the problem yet. The first stage is to define the
problem.

I need
opinions here.


Anyone can give you opinions. I think the best ice cream in the world is
plum flavour. Others may differ. Is this useful information? Probably
not. That's the thing about opinions.

Treat this as if it were your own experiment and share your approach or
theory.


What am I trying to prove or disprove with this theory? Do I care about
rate conversion artifacts, ultrasonic perception, converter artifacts, or
something else? What am I trying to prove the audibility of?

Until you frame the question better, you can't learn anything


Then please tell me how you would approach the question.


We don't know what the question IS.

My recommendation, FWIW, is to begin by asking what, specifically,
is different in a modern recording-and-reproduction chain between
different "sampling rates". Defining "sampling rates" is the first
hurdle. It's *not* the simple number that your post implies, (and
Scott has strongly implied the solution).


It doesn't have to be complicated. It could be tedious, but not complicated.
First, we use the same brand DA or AD converter for everything. Let's assume
that I could manage to get a manufacturer to match several units to an
acceptable level. (Yes, we need to define what would be an acceptable error
between the different units)

Then, we have to decide if the following methods are acceptable:

A-downsample a sound file to different rates and then upsample back the
resultiung sound files to play everything at the same rate.
B-downsample and then play every file at its respective downsampled rate.
C-or use X number of computers (one for each sample rate to be tested),
record with the same brand of AD, feed a live performance to every computer
at the same time and those computer would each record at a different
sampling rate. And then play the files back at their native sample rate.
D-Any method someone can think of around here.

The problem with C is, what do I record, and how do I record it to make sure
that the source will be "Accepted" by the audio community as a "Benchmark"
recording?


It depends on what you're trying to measure. These procedures measure
different things. You need to know what to measure before you make a
measurement.

Scott and Paul gave good ideas, I would like to have more. I need to hear
all the potential issues and questions. Then, I will be able to design a
proper experiment.


You need a hypothesis first.

I first need to isolate the "Sampling Rate" thing . Then, investigate the
different elements in the chain.

For example, how will a mastering engineer approach a song, which would have
to be delivered at different rates (CD, SACD, whatever)


Depends on a lot of factors. SACD isn't just a different sampling rate, it
is a totally different encoding system. Many mastering facilities are using
analogue chains anyway, so they can encode with any format at any rate off
the analogue feed on the fly.

Then when a end user listens to the product, what are the different elements
in the " live source-recording-mixing-mastering-delivery medium-end user
playback unit" chain.


There are lots of them.

But first I need to isolate the sampling rate issue and decide which
processing artifacts are acceptable in such a test.


What issue? We still don't know what issue you are talking about.
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."
  #14   Report Post  
Posted to rec.audio.pro
Chris Hornbeck Chris Hornbeck is offline
external usenet poster
 
Posts: 1,744
Default Sample rate conversion question

On Fri, 15 Aug 2008 20:55:55 -0400, "karle"
wrote:

Thanks for the answers. I would appreciate it if people contributed to the
problem instead raising questions without trying to answer those. I need
opinions here.


I disagree. IMO the framing of questions is the *only* important
thing. Answers are trivial bull****. Hey! Just like this one!

Treat this as if it were your own experiment and share your approach or
theory.

Until you frame the question better, you can't learn anything


Then please tell me how you would approach the question.


Cool. My approach to this, a topic that greatly interests me
BTW, is to look at the block diagram of the whole recording
and reproduction process. What blocks go into this diagram,
and how are they arranged?

A sufficiently interesting-fer-starters block diagram has just
a single linear chain from input signal to output signal. Call
it microphone to loudspeaker, to broaden the assumptions. Plenty
close enough for a conceptual start.

Within this linear chain are *multiple* conversions to-'n-from
various sampling rates and bit depths. This is the interesting
part (along with basic sampling theory).

Scott hinted at the missing link: " Note that most converters today
are sigma-delta converters so you already have some
internal rate conversion going on inside the box."


The non-trivial question is going to be related to something
very much other than "sampling rate" as expressed in an advert.
What *exactly* about the topic is interesting? That's the
non-trivial question.

All the best fortune,
Chris Hornbeck
  #15   Report Post  
Posted to rec.audio.pro
jwvm jwvm is offline
external usenet poster
 
Posts: 336
Default Sample rate conversion question

On Aug 15, 6:04*pm, Mike Rivers wrote:

snip

I can answer that right now. No. The average people are very
non-discriminatory. Audio fussbudjits, however, can hear the difference
between 44.1 and 44.1 and 44.1 kHz all of the same word length.


Hey, those guys are so good they can hear the difference between
wooden and plastic knobs on volume controls. They can even tell when a
CD has been demagnetized!




  #16   Report Post  
Posted to rec.audio.pro
jwvm jwvm is offline
external usenet poster
 
Posts: 336
Default Sample rate conversion question

On Aug 15, 8:15*pm, (Scott Dorsey) wrote:
In article m3npk.28$5C.0@trnddc02, Mike Rivers wrote:



snip
The average person can't tell the difference between CD and a cassette
that has been dunked in orange soda.


Sorry Scott but I have to disagree with you there. Most people will
immediately tell you that the cassette tastes much better.
  #17   Report Post  
Posted to rec.audio.pro
Chris Hornbeck Chris Hornbeck is offline
external usenet poster
 
Posts: 1,744
Default Sample rate conversion question

On Fri, 15 Aug 2008 20:29:32 -0700 (PDT), jwvm wrote:

On Aug 15, 6:04*pm, Mike Rivers wrote:
Audio fussbudjits, however, can hear the difference
between 44.1 and 44.1 and 44.1 kHz all of the same word length.


Hey, those guys are so good they can hear the difference between
wooden and plastic knobs on volume controls. They can even tell when a
CD has been demagnetized!


That ain't no thang. I can personally hear if a CD is played
upside down. Without looking!, let me add.

So there,
Chris Hornbeck
  #18   Report Post  
Posted to rec.audio.pro
karle karle is offline
external usenet poster
 
Posts: 4
Default Sample rate conversion question

Ok I will re-explain in detail then.

What I want to do is:

1- Record a performance with 2 microphones fed into a console
2-Take the console outputs and send these to 7 identical recorders (computer
/ AD converter, or maybe standalone recorders), each recording at a
different sampling rate
3- Play back the various recordings to test subjects to see which recordings
they prefer
4- Then run the experiment again with X brands of converters, but each brand
recording at only at the most preferred and the least preferred sample rate.
5- If, among all the brands, the most preferred sample rate remains, then I
can assume that the distortions and artifacts introduced by the converters
aren't as meaningful as the number of samples recorded per second. Because
the filter slopes and converter performance will vary from manufacturer to
manufacturer.
I will then concentrate my efforts on the ultrasonic perception part.
6-If, among all the brands, the most preferred sample rate differs, then I
need to dig deeper.

What are the flaws in this plan?

My main goal is to establish a transparency "upper bound"
sampling rate, as Bob stated it. If I reach point 6 in my above plan, then I
will try to isolate the various elements.




"Chris Hornbeck" wrote in message
...
On Fri, 15 Aug 2008 20:55:55 -0400, "karle"
wrote:

Thanks for the answers. I would appreciate it if people contributed to the
problem instead raising questions without trying to answer those. I need
opinions here.


I disagree. IMO the framing of questions is the *only* important
thing. Answers are trivial bull****. Hey! Just like this one!

Treat this as if it were your own experiment and share your approach or
theory.

Until you frame the question better, you can't learn anything


Then please tell me how you would approach the question.


Cool. My approach to this, a topic that greatly interests me
BTW, is to look at the block diagram of the whole recording
and reproduction process. What blocks go into this diagram,
and how are they arranged?

A sufficiently interesting-fer-starters block diagram has just
a single linear chain from input signal to output signal. Call
it microphone to loudspeaker, to broaden the assumptions. Plenty
close enough for a conceptual start.

Within this linear chain are *multiple* conversions to-'n-from
various sampling rates and bit depths. This is the interesting
part (along with basic sampling theory).

Scott hinted at the missing link: " Note that most converters today
are sigma-delta converters so you already have some
internal rate conversion going on inside the box."


The non-trivial question is going to be related to something
very much other than "sampling rate" as expressed in an advert.
What *exactly* about the topic is interesting? That's the
non-trivial question.

All the best fortune,
Chris Hornbeck



  #19   Report Post  
Posted to rec.audio.pro
Ty Ford Ty Ford is offline
external usenet poster
 
Posts: 2,287
Default Sample rate conversion question

On Fri, 15 Aug 2008 08:54:34 -0400, karle wrote
(in article ):

Hi All,

I will be doing some tests, to find out if people can really hear the
difference if a song is played at different sampling rates.

Does anyone knows about a computer / Software / Hardware bundle that would
enable playing songs and changing the master clock on the fly? I would need
the DA converter to switch its clock automatically, according to the signal
it receives at the inputs and the finest equipment if possible.

Also, if you already conducted such an experiment, I'd like to know how you
did it.

Thanks,

Karle
McGill University


Karle,

I don't know what sample rates you're looking to test, but I can hear the
difference between 44.1 and 48 kHz.

Regards,

Ty Ford



--Audio Equipment Reviews Audio Production Services
Acting and Voiceover Demos http://www.tyford.com
Guitar player?:http://www.youtube.com/watch?v=4RZJ9MptZmU

  #20   Report Post  
Posted to rec.audio.pro
Mike Rivers Mike Rivers is offline
external usenet poster
 
Posts: 8,744
Default Sample rate conversion question

karle wrote:

What I want to do is:
1- Record a performance with 2 microphones fed into a console
2-Take the console outputs and send these to 7 identical recorders (computer
/ AD converter, or maybe standalone recorders), each recording at a
different sampling rate


What are the flaws in this plan?


I don't think you can find seven identical recorders or A/D converters
that span that wide a range of sample rates. Did you have some in mind,
or is that what you're asking about?

Really high end converters that will do 192 kHz with good quality don't
go any lower than 44.1 kHz, and crummy converters that can do 192 kHz
are just plain crummy converters. I don't see much point in using the
near-intermediate rates - for example, 44.1 and 48 kHz. So if you
conduct your experiment at the "standard" rates of 44.1 or 48 kHz (take
your pick), 96 kHz, and 192 kHz, you should cover the range for your
experiment pretty well. Ask your friendly local Apogee dealer to lend
you three Rosetta converters. The trick, though, will be to
simultaneously record the digital outputs of the three different sample
rates to the same recorder or computer.

Since your goal is to test converters (I think), perhaps a better
approach than using microphones into multiple converters is to use a
standard test recording and record one pass of that into each converter.
Without very good microphones in a very good environment, it isn't going
to be any better than recording the microphone first, the best way you
can (or use a commercial recording). That may seem like it will defeat
part of your test plan, but it may be the best you can do unless you
have a very wealthy sponsor.





--
If you e-mail me and it bounces, use your secret decoder ring and reach
me he
double-m-eleven-double-zero at yahoo -- I'm really Mike Rivers
)


  #21   Report Post  
Posted to rec.audio.pro
Scott Dorsey Scott Dorsey is offline
external usenet poster
 
Posts: 16,853
Default Sample rate conversion question

karle wrote:
Ok I will re-explain in detail then.

What I want to do is:

1- Record a performance with 2 microphones fed into a console
2-Take the console outputs and send these to 7 identical recorders (computer
/ AD converter, or maybe standalone recorders), each recording at a
different sampling rate
3- Play back the various recordings to test subjects to see which recordings
they prefer
4- Then run the experiment again with X brands of converters, but each brand
recording at only at the most preferred and the least preferred sample rate.
5- If, among all the brands, the most preferred sample rate remains, then I
can assume that the distortions and artifacts introduced by the converters
aren't as meaningful as the number of samples recorded per second. Because
the filter slopes and converter performance will vary from manufacturer to
manufacturer.
I will then concentrate my efforts on the ultrasonic perception part.
6-If, among all the brands, the most preferred sample rate differs, then I
need to dig deeper.

What are the flaws in this plan?


You still don't know what you're measuring when you do it. You have
preferences but you have no idea _why_ anyone has a preference.

See... in a perfect world, the ONLY difference between different sample
rates is due to bandlimiting. If you use a higher sample rate, the
frequency response is extended.

But, we don't live in a perfect world, so sometimes there are artifacts
that are not due to bandlimiting. For example, with the Panasonic SV-3700,
there is a dramatic difference in sound quality between 44.1 and 48 ksamp/sec
operation. That's not due to the change in sample rates itself, per se,
because it's not due to the bandlimiting. It's because the converters in
the SV-3700 were designed by idiots and have massive aliasing problems,
and those aliasing problems change with the sample rate.

On the other hand, you can take a Prism AD-124, sold at about the same
time as the SV-3700, which was designed by people who knew what they were
doing, and really hear no difference between the 44.1 and 48 ksamp/sec
rates.

So.... do you want to change sample rates in order to tell if people can
hear bandlimiting, or do you want to change sample rates in order to tell
if the converters have other artifacts? Because both of these things
happen at the same time.

My main goal is to establish a transparency "upper bound"
sampling rate, as Bob stated it. If I reach point 6 in my above plan, then I
will try to isolate the various elements.


Folks have done testing for years and years to determing "upper bound"
sampling rates in the real world, and it has changed as the converter
technology has changed. The old DASH recorders ran at 54 ksamp/sec, because
with the converter technology of the day that was the lowest rate they
could run at without clearly audible high frequency problems. Today the
converter technology is much better, and so we can run at much lower rates
without audible differences.

So if you test YOUR equipment with a listening panel, and you record and
play at different rates, the results you get are valid ONLY for YOUR equipment.
They cannot be generallized. If you want a test that can be generallized
you have to be a lot more sophisticated about it. If that's all you care
about, that's fine, but strictly speaking it's not very useful.

Check the test in the recent JAES issue that folks have referred to. It
wasn't last quarter, I think it was the issue or two before that. It was
nicely one.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."
  #22   Report Post  
Posted to rec.audio.pro
Scott Dorsey Scott Dorsey is offline
external usenet poster
 
Posts: 16,853
Default Sample rate conversion question

Ty Ford wrote:

I don't know what sample rates you're looking to test, but I can hear the
difference between 44.1 and 48 kHz.


With what converters?

I'd be surprised if you could hear it with a good modern sigma-delta
converter. I'd be worried if you couldn't hear it with a Panasonic
DAT machine.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."
  #23   Report Post  
Posted to rec.audio.pro
hank alrich hank alrich is offline
external usenet poster
 
Posts: 4,736
Default Sample rate conversion question

Mike Rivers wrote:

karle wrote:

What I want to do is:
1- Record a performance with 2 microphones fed into a console
2-Take the console outputs and send these to 7 identical recorders (computer
/ AD converter, or maybe standalone recorders), each recording at a
different sampling rate


What are the flaws in this plan?


I don't think you can find seven identical recorders or A/D converters
that span that wide a range of sample rates. Did you have some in mind,
or is that what you're asking about?


We've had discussions about how this or that convertor may or may not be
optimized for premier performance at any particualr sample rate.

snip


--
ha
Iraq is Arabic for Vietnam
  #24   Report Post  
Posted to rec.audio.pro
Richard Crowley Richard Crowley is offline
external usenet poster
 
Posts: 4,172
Default Sample rate conversion question

"Scott Dorsey" wrote ...
Check the test in the recent JAES issue that folks have referred to. It
wasn't last quarter, I think it was the issue or two before that. It was
nicely one.


One could even argue that the OP's experiment has already been
done (conclusively, likely many times, even), but the OP hasn't
done enough of his own research to realize that. Or at best, he
hasn't adequately explained how his proposed experiment
differentiates from previous published works.


  #25   Report Post  
Posted to rec.audio.pro
Mike Rivers Mike Rivers is offline
external usenet poster
 
Posts: 8,744
Default Sample rate conversion question

Ty Ford wrote:

I don't know what sample rates you're looking to test, but I can hear the
difference between 44.1 and 48 kHz.


Hearing a difference isn't hard. I can hear a difference between today
and yesterday. But is there something about what you hear between 44.1
and 48 kHz that would lead you to use one of those sample rates all the
time if you had a choice?

In the DAT days, there were some pretty poor implementations of
filtering that exaggerated the difference between the two sample rate
settings, but the difference wasn't because of more samples per
unit time, it was how they were sampled or reconstructed. But you
probably knew that.

--
If you e-mail me and it bounces, use your secret decoder ring and reach
me he
double-m-eleven-double-zero at yahoo -- I'm really Mike Rivers
)


  #26   Report Post  
Posted to rec.audio.pro
Scott Dorsey Scott Dorsey is offline
external usenet poster
 
Posts: 16,853
Default Sample rate conversion question

In article ,
Richard Crowley wrote:
"Scott Dorsey" wrote ...
Check the test in the recent JAES issue that folks have referred to. It
wasn't last quarter, I think it was the issue or two before that. It was
nicely one.


One could even argue that the OP's experiment has already been
done (conclusively, likely many times, even), but the OP hasn't
done enough of his own research to realize that. Or at best, he
hasn't adequately explained how his proposed experiment
differentiates from previous published works.


Oh, pretty much all of the possible experiments (and there are hundreds)
that meet the OP's description have been done. Some of them have been
done better than others, though. But that doesn't mean they aren't worth
doing again (possibly with better control).
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
  #27   Report Post  
Posted to rec.audio.pro
Mark Mark is offline
external usenet poster
 
Posts: 966
Default Sample rate conversion question

On Aug 16, 3:07*pm, (Scott Dorsey) wrote:
In article ,

Richard Crowley wrote:
"Scott Dorsey" wrote ...
Check the test in the recent JAES issue that folks have referred to. *It
wasn't last quarter, I think it was the issue or two before that. *It was
nicely one.


One could even argue that the OP's experiment has already been
done (conclusively, likely many times, even), but the OP hasn't
done enough of his own research to realize that. *Or at best, he
hasn't adequately explained how his proposed experiment
differentiates from previous published works.


Oh, pretty much all of the possible experiments (and there are hundreds)
that meet the OP's description have been done. *Some of them have been
done better than others, though. *But that doesn't mean they aren't worth
doing again (possibly with better control).
--scott
--
"C'est un Nagra. *C'est suisse, et tres, tres precis."



It disappoints me that after all this discussion, no one mentions
making a MEASUREMENT instead of a listening test.

http://src.infinitewave.ca/

Mark

  #28   Report Post  
Posted to rec.audio.pro
hank alrich hank alrich is offline
external usenet poster
 
Posts: 4,736
Default Sample rate conversion question

Mark wrote:

On Aug 16, 3:07 pm, (Scott Dorsey) wrote:
In article ,

Richard Crowley wrote:
"Scott Dorsey" wrote ...
Check the test in the recent JAES issue that folks have referred to. It
wasn't last quarter, I think it was the issue or two before that. It was
nicely one.


One could even argue that the OP's experiment has already been
done (conclusively, likely many times, even), but the OP hasn't
done enough of his own research to realize that. Or at best, he
hasn't adequately explained how his proposed experiment
differentiates from previous published works.


Oh, pretty much all of the possible experiments (and there are hundreds)
that meet the OP's description have been done. Some of them have been
done better than others, though. But that doesn't mean they aren't worth
doing again (possibly with better control).
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."



It disappoints me that after all this discussion, no one mentions
making a MEASUREMENT instead of a listening test.

http://src.infinitewave.ca/

Mark


Probably because few of us can hear a measurement. Those who can
disagree on the signficance of their experience.

--
ha
Iraq is Arabic for Vietnam
  #29   Report Post  
Posted to rec.audio.pro
Scott Dorsey Scott Dorsey is offline
external usenet poster
 
Posts: 16,853
Default Sample rate conversion question

Mark wrote:

It disappoints me that after all this discussion, no one mentions
making a MEASUREMENT instead of a listening test.

http://src.infinitewave.ca/


That's the SECOND step; once you know what is audible you do the measurements
to see how much of whatever is measurable is audible.

The measurements aren't of much use without listening tests, but it's true
the listening tests aren't of much use without measurements either.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
  #30   Report Post  
Posted to rec.audio.pro
William Sommerwerck William Sommerwerck is offline
external usenet poster
 
Posts: 4,718
Default Sample rate conversion question

It disappoints me that after all this discussion, no one mentions
making a MEASUREMENT instead of a listening test.


Audio equipment is supposed to be listened to, not just measured.
Measurement are largely meaningless unless they have been correlated with
_reliable_, _meaningful_ listening tests.

Pace, Arny...




  #31   Report Post  
Posted to rec.audio.pro
Chris Hornbeck Chris Hornbeck is offline
external usenet poster
 
Posts: 1,744
Default Sample rate conversion question

On Sat, 16 Aug 2008 18:02:47 -0700 (PDT), Mark
wrote:

It disappoints me that after all this discussion, no one mentions
making a MEASUREMENT instead of a listening test.


If you can get the OP to decide what to measure, or can
convince him to measure what *you* consider important to
measure, you'll be able to add a data point.

But without defining the "sample rate conversion question"
of the title, it'll still be all smoke, no buzz.

Nebulous "questions" without any real preparation are
worse than useless. Real questions are important, and it's
especially important to protect real questions from the
distractions of bogus "questions".

We all have to start somewhere, and IMpersonalE getting
something really and publicly wrong is sometimes a good
starting point. That's my recommendation to the OP. FWIW,
probably not much.

Much thanks, as always,
Chris Hornbeck
  #32   Report Post  
Posted to rec.audio.pro
Arny Krueger Arny Krueger is offline
external usenet poster
 
Posts: 17,262
Default Sample rate conversion question

"karle" wrote in message

Thanks,

I want to know if average people can really hear the
difference between, 44.1, 48, 88.2, 96, an 192 khz.


The advice I need is, should I:


1-Start with a 192khz source, downsample it and play it
back at the downsampled rates on the D to A converter


A waste of time.

2-Start with a 192khz source, downsample it and play it
back at the highest rate on the D to A converter


http://www.aes.org/e-lib/browse.cfm?elib=14195

Claims both published and anecdotal are regularly made for audibly superior
sound quality for two-channel audio encoded with longer word lengths and/or
at higher sampling rates than the 16-bit/44.1-kHz CD standard. The authors
report on a series of double-blind tests comparing the analog output of
high-resolution players playing high-resolution recordings with the same
signal passed through a 16-bit/44.1-kHz “bottleneck.” The tests were
conducted for over a year using different systems and a variety of subjects.
The systems included expensive professional monitors and one high-end system
with electrostatic loudspeakers and expensive components and cables. The
subjects included professional recording engineers, students in a university
recording program, and dedicated audiophiles. The test results show that the
CD-quality A/D/A loop was undetectable at normal-to-loud listening levels,
by any of the subjects, on any of the playback systems. The noise of the
CD-quality loop was audible only at very elevated levels.

Discussed here.

http://www.hydrogenaudio.org/forums/...howtopic=57406

3-Record the same performance at different sample rates
and then play it back at its native sample rate


If ny native sample rate you mean 44.1 or 48 kHz, this is the thinking
man's solution.



  #33   Report Post  
Posted to rec.audio.pro
Ty Ford Ty Ford is offline
external usenet poster
 
Posts: 2,287
Default Sample rate conversion question

On Sat, 16 Aug 2008 14:56:56 -0400, Mike Rivers wrote
(in article YpFpk.54$UX.24@trnddc03):

Ty Ford wrote:

I don't know what sample rates you're looking to test, but I can hear the
difference between 44.1 and 48 kHz.


Hearing a difference isn't hard. I can hear a difference between today
and yesterday. But is there something about what you hear between 44.1
and 48 kHz that would lead you to use one of those sample rates all the
time if you had a choice?

In the DAT days, there were some pretty poor implementations of
filtering that exaggerated the difference between the two sample rate
settings, but the difference wasn't because of more samples per
unit time, it was how they were sampled or reconstructed. But you
probably knew that.



I did the test once before when someone asked how much difference there was.
I forget when exactly, so it was probably DAT at 16-bit. The difference I
heard was more "air" (I sort of dislike the term) and HF extension with the
48 kHz.

I remember thinking hmm, a little more top end at 48kHz. Too bad that won't
make it to CD, but it wasn't a deal killer.

Regards,

Ty Ford


--Audio Equipment Reviews Audio Production Services
Acting and Voiceover Demos http://www.tyford.com
Guitar player?:http://www.youtube.com/watch?v=4RZJ9MptZmU

  #34   Report Post  
Posted to rec.audio.pro
Chris Hornbeck Chris Hornbeck is offline
external usenet poster
 
Posts: 1,744
Default Sample rate conversion question

On Fri, 15 Aug 2008 21:31:07 -0400, "Soundhaspriority"
wrote:

However, I seem to recall, and perhaps Scott or someone else can refresh my
memory, that someone did reasonably good quality research that established
something like a 60k sampling frequency, and 22 bits, as required for
ultimate transparency. It would be interesting to see how that author
handled the confounding variables.


Chapter 18 of Bob Katz's book _Mastering Audio_ is devoted to this
topic, and references some personal correspondence with Jim
Johnston - the original horse's mouth - and Tom Stockham's
work for the Soundstream project.

Highly recommended, for this and many other reasons. Very accessable
for folks with a deep general knowledge seeking specifics. Also,
very well written. A class act all the way around.

Much thanks, as always,
Chris Hornbeck
  #35   Report Post  
Posted to rec.audio.pro
[email protected] really@ao1.com is offline
external usenet poster
 
Posts: 1
Default Sample rate conversion question

"karle" wrote:

I want to know if average people can really hear the difference between,
44.1, 48, 88.2, 96, an 192 khz.



Oh, "average people".

Average people can't hear much at all. It you want to stratify THEM, use
rates like 22050 vs 11025.

I think you're in academic fantasy land.


  #36   Report Post  
Posted to rec.audio.pro
David Morgan \(MAMS\) David Morgan \(MAMS\) is offline
external usenet poster
 
Posts: 1,222
Default Sample rate conversion question


"karle" wrote:

I want to know if average people can really hear the difference between,
44.1, 48, 88.2, 96, an 192 khz.



Anything above 48K is little more than vanity.... 88.2 makes for good
math.... and anything greater than 88.2 is wasted drive space and
serves only to put money into the pockets of marketing specialists.

JMHO, of course....

DM







--
David Morgan (MAMS)
Morgan Audio Media Service
http://www.m-a-m-s DOT com
Dallas, Texas (214) 662-9901
_____________________________
http://www.januarysound.com



  #37   Report Post  
Posted to rec.audio.pro
Mike Rivers Mike Rivers is offline
external usenet poster
 
Posts: 8,744
Default Sample rate conversion question

David Morgan (MAMS) wrote:

.... 88.2 makes for good math....


Only in arithmetic. It's really no easier to convert 88.2 kHz to 44.1
kHz as to convert 96 kHz, or 50 kHz or 163.4 kHz. You still have to
recalculate every sample. You can't get good results by simply taking
every other sample as some people infer.



--
If you e-mail me and it bounces, use your secret decoder ring and reach
me he
double-m-eleven-double-zero at yahoo -- I'm really Mike Rivers
)
  #38   Report Post  
Posted to rec.audio.pro
Scott Dorsey Scott Dorsey is offline
external usenet poster
 
Posts: 16,853
Default Sample rate conversion question

Chris Hornbeck wrote:
On Fri, 15 Aug 2008 21:31:07 -0400, "Soundhaspriority"
wrote:

However, I seem to recall, and perhaps Scott or someone else can refresh my
memory, that someone did reasonably good quality research that established
something like a 60k sampling frequency, and 22 bits, as required for
ultimate transparency. It would be interesting to see how that author
handled the confounding variables.


Chapter 18 of Bob Katz's book _Mastering Audio_ is devoted to this
topic, and references some personal correspondence with Jim
Johnston - the original horse's mouth - and Tom Stockham's
work for the Soundstream project.

Highly recommended, for this and many other reasons. Very accessable
for folks with a deep general knowledge seeking specifics. Also,
very well written. A class act all the way around.


Yes, absolutely! And what is interesting about this is that it talks
somewhat about HOW the technology affects the required minimum sampling rate.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
  #39   Report Post  
Posted to rec.audio.pro
Scott Dorsey Scott Dorsey is offline
external usenet poster
 
Posts: 16,853
Default Sample rate conversion question

Mike Rivers wrote:
David Morgan (MAMS) wrote:

.... 88.2 makes for good math....


Only in arithmetic. It's really no easier to convert 88.2 kHz to 44.1
kHz as to convert 96 kHz, or 50 kHz or 163.4 kHz. You still have to
recalculate every sample. You can't get good results by simply taking
every other sample as some people infer.


You can, IF you do a lowpass at 22 KHz first. This is called "filtering
and decimating" and is sort of the crudest possible sample rate conversion.

Most actual realworld software, though, uses the same algorithm whether or
not you are dividing by an integral factor.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
  #40   Report Post  
Posted to rec.audio.pro
David Morgan \(MAMS\) David Morgan \(MAMS\) is offline
external usenet poster
 
Posts: 1,222
Default Sample rate conversion question


"Scott Dorsey" wrote in message ...
Mike Rivers wrote:
David Morgan (MAMS) wrote:

.... 88.2 makes for good math....


Only in arithmetic. It's really no easier to convert 88.2 kHz to 44.1
kHz as to convert 96 kHz, or 50 kHz or 163.4 kHz. You still have to
recalculate every sample. You can't get good results by simply taking
every other sample as some people infer.


You can, IF you do a lowpass at 22 KHz first. This is called "filtering
and decimating" and is sort of the crudest possible sample rate conversion.

Most actual realworld software, though, uses the same algorithm whether or
not you are dividing by an integral factor.


Like I said.... "makes for good math".... I didn't mention audio. ;-)

I still record all of my sessions at 24/44.1, and I'm on Digi 192s / Apogee.

Dither sucks. (Again, JMHO).


DM



Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Adobe Audition 1.5 allows WMA monoaural audio at 44.1 KHz sample-rate with a bit-rate of 20 kbps Radium[_4_] Audio Opinions 13 July 23rd 07 09:45 PM
question on sample rate (and conversion etc) MisterE Pro Audio 13 April 5th 06 06:07 AM
Help wanted: 2 channel 24bit AES/EBU sample rate conversion 96khz to 48khz Marcel de Velde Pro Audio 9 March 13th 04 09:30 AM
USB digital transfer sample rate question Anders J Tech 2 September 3rd 03 09:15 PM
USB digital transfer sample rate question Anders J High End Audio 1 September 2nd 03 05:57 PM


All times are GMT +1. The time now is 12:07 PM.

Powered by: vBulletin
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 AudioBanter.com.
The comments are property of their posters.
 

About Us

"It's about Audio and hi-fi"