[[email protected]]

I am trying to track down a citation to Julian Hirsch and his
comments
on digital audio. Specifically he is supposed to have opined (in 1972
or thereabouts) that a 5% sample of an analog sound signal would be
sufficient to be indistinguishable from the original. This is
supposedly based on some experiments but I can find no reference to
these or this statement anywhere. I would wonder what kind of digital
audio encoders/decoders would even be available in 1972 for this
assertion to be made.

Can anyone point me in the right direction.


Cheers


Jim

[[email protected]]

On Jan 28, 2:53 pm, " wrote:
I am trying to track down a citation to Julian Hirsch and his
comments
on digital audio. Specifically he is supposed to have opined (in 1972
or thereabouts) that a 5% sample of an analog sound signal would be
sufficient to be indistinguishable from the original. This is
supposedly based on some experiments but I can find no reference to
these or this statement anywhere. I would wonder what kind of digital
audio encoders/decoders would even be available in 1972 for this
assertion to be made.

The "5% sample" comment has little to do with "digital
sampling" in the sense you are assuming.

Instead, consider that stereo recording and playback
technology is missing SO MUCH of the original sound
field and yet is so satisfying to so many people. Taking
a sound field which is a multii-dimensional (in the
mathematical sense) and reducing it to two time-variant
variations in voltage simply throws away the vast majority
of the directional and temporal information. It's not
a "sampling" issue in the sense of discrete-time sampling.

Now, bring the "discrete time sampling" issue back into
play: Shannon showed over a half century ago that a discrete
time sampled representation of that electrical signal can
capture 100% of it.

If there is a flaw in our recording technology, and we're only
capturing 5% of the music, it's not in the digital realm we're
losing the other 95%, it's that conspiracy of two channels
playing back over two speakers and robbinb the human of
the abolity to be immersed in a true replica of ths sound field
that said human has a chance of properly "sampling."

This despite the rabid and largely uninformed rantings of
a number of alleged "high-end experts."

[Dirk Bruere at NeoPax]

wrote:

On Jan 28, 2:53 pm, " wrote:
I am trying to track down a citation to Julian Hirsch and his
comments
on digital audio. Specifically he is supposed to have opined (in 1972
or thereabouts) that a 5% sample of an analog sound signal would be
sufficient to be indistinguishable from the original. This is
supposedly based on some experiments but I can find no reference to
these or this statement anywhere. I would wonder what kind of digital
audio encoders/decoders would even be available in 1972 for this
assertion to be made.

The "5% sample" comment has little to do with "digital
sampling" in the sense you are assuming.

Instead, consider that stereo recording and playback
technology is missing SO MUCH of the original sound
field and yet is so satisfying to so many people. Taking
a sound field which is a multii-dimensional (in the
mathematical sense) and reducing it to two time-variant
variations in voltage simply throws away the vast majority
of the directional and temporal information. It's not
a "sampling" issue in the sense of discrete-time sampling.

Now, bring the "discrete time sampling" issue back into
play: Shannon showed over a half century ago that a discrete
time sampled representation of that electrical signal can
capture 100% of it.

If there is a flaw in our recording technology, and we're only
capturing 5% of the music, it's not in the digital realm we're
losing the other 95%, it's that conspiracy of two channels
playing back over two speakers and robbinb the human of
the abolity to be immersed in a true replica of ths sound field
that said human has a chance of properly "sampling."

This despite the rabid and largely uninformed rantings of
a number of alleged "high-end experts."

Unfortunately we have only two ears with which to appreciate that lost 95%

--
Dirk

http://www.onetribe.me.uk - The UK's only occult talk show
Presented by Dirk Bruere and Marc Power on ResonanceFM 104.4
http://www.resonancefm.com

[Arny Krueger]

wrote in message
oups.com
I am trying to track down a citation to Julian Hirsch and
his comments on digital audio.

Specifically he is supposed to have
opined (in 1972 or thereabouts) that a 5% sample of an
analog sound signal would be sufficient to be
indistinguishable from the original. This is supposedly
based on some experiments but I can find no reference to
these or this statement anywhere. I would wonder what
kind of digital audio encoders/decoders would even be
available in 1972 for this assertion to be made.

Can anyone point me in the right direction.

I'm kinda surprised by this because Julian wasn't that far from being right,
if you consider the statistics related to perceptual coding. A 75 Kbps mono
file represents approximately 5% of the data in a 44/16 channel. At the
current SOTA the best perceptual coders don't miss this by a whole lot.

However, I'm under the impression that we did not know enough about
perceptual coding to reasonably make this kind of an estimate in 1972. 1992
would be more like it. In 1972 masking was known to exist, but AFAIK its
potential was not well known enough to make a good estimate. Maybe Julian
was lucky.

[Barry Mann]

In , on 01/28/07
at 08:48 PM, Dirk Bruere at NeoPax said:

[ ... ]

Unfortunately we have only two ears with which to appreciate that lost
95%

When immersed in a concert setting we can easily feel the bass content
and view and smell other happy listeners. Some of that "sound" will
travel in different materials at different speeds. You may end up
feeling the bass in your feet before you can hear the direct sound.

-----------------------------------------------------------
spam:
wordgame:123(abc):14 9 20 5 2 9 18 4 at 22 15 9 3 5 14 5 20 dot 3 15
13 (Barry Mann)
[sorry about the puzzle, spammers are ruining my mailbox]
-----------------------------------------------------------

[[email protected]]

In 1972 masking was known to exist, but AFAIK its
potential was not well known enough to make a good estimate. MaybeJulian
was lucky.

Or maybe he never said such a thing and what we have here is an urban
legend. I have searched high and low and all I have so far is a chap
from headfi.org saying he remembers seeing this assertion in print
(possibly in 1983 not 1972 as originally stated - sorry ) but I would
be interested in knowing is how such a figure could be derived. How
would you go about quantifying the data that exists in an audio signal
- i.e unrecorded - in bit terms before you capture it ?

[Trevor Wilson]

"Barry Mann" wrote in message
.. .
In , on 01/28/07
at 08:48 PM, Dirk Bruere at NeoPax said:

[ ... ]

Unfortunately we have only two ears with which to appreciate that lost
95%

When immersed in a concert setting we can easily feel the bass content
and view and smell other happy listeners. Some of that "sound" will
travel in different materials at different speeds. You may end up
feeling the bass in your feet before you can hear the direct sound.

**A bibaural recording, with the microphones appropriately coupled to the
floor, should restore most of that issue. I've heard some VERY convincing
binaural recordings. Two channels which make 5.1 channels sound like crap.


--
Trevor Wilson
www.rageaudio.com.au




--
Posted via a free Usenet account from http://www.teranews.com

[Arny Krueger]

wrote in message
ps.com
In 1972 masking was known to exist, but AFAIK its
potential was not well known enough to make a good
estimate. MaybeJulian was lucky.

Or maybe he never said such a thing and what we have here
is an urban legend. I have searched high and low and all
I have so far is a chap from headfi.org saying he
remembers seeing this assertion in print (possibly in
1983 not 1972 as originally stated - sorry ) but I would
be interested in knowing is how such a figure could be
derived.

1983 is more believable. It is still a little early.

How would you go about quantifying the data that
exists in an audio signal - i.e unrecorded - in bit terms
before you capture it ?

The means for characterizing the amount of information in a channel (whether
analog or digital) has been known for at least 50 years - ever since Claude
Shannon's ground-breaking work at Bell Labs in 1948-49. The key parameters
are bandwidth and dynamic range.

[Mark & Mary Ann Weiss]

If there is a flaw in our recording technology, and we're only
capturing 5% of the music, it's not in the digital realm we're
losing the other 95%, it's that conspiracy of two channels
playing back over two speakers and robbinb the human of
the abolity to be immersed in a true replica of ths sound field
that said human has a chance of properly "sampling."

This despite the rabid and largely uninformed rantings of
a number of alleged "high-end experts."

Unfortunately we have only two ears with which to appreciate that lost 95%

The problem is to recreate the right number of sound arrivals to each ear,
and that's where conventional stereo introduces double sets of sound
arrivals, which do not replicate the original sound.
Ie, a solo instrument plays through stereo speakers. Each speaker produces
two sound arrivals for the ears--one for left and one for right ear times
two speakers.
In a live performance, there would be only one set of sound arrivals, one
for left and right ears, each.
Bob Carver, in 1981, tackled this problem with cross-fed interference
signals, to cancel the extra unwanted sound arrivals. It works very well in
my setup. Although you have to be sitting in the center line of the speakers
and speaker design has to follow a certain set of parameters for this to
work effectively.


--


Take care,



Mark & Mary Ann Weiss



VIDEO PRODUCTION . FILM SCANNING . DVD MASTERING . AUDIO RESTORATION
Hear my Kurzweil Creations at: www.dv-clips.com/theater.htm
www.basspig.com The Bass Pig's Lair - 15,000 Watts of Driving Stereo!
Business sites at:
www.mwcomms.com
www.adventuresinanimemusic.com
-

[[email protected]]

The means for characterizing the amount of information in a channel (whether
analog or digital) has been known for at least 50 years - ever since Claude
Shannon's ground-breaking work at Bell Labs in 1948-49. The key parameters
are bandwidth and dynamic range.

Ah, so if we use 20 - 20,000 hz and 96db then we get ~16 bits ?

[Dirk Bruere at NeoPax]

Mark & Mary Ann Weiss wrote:
If there is a flaw in our recording technology, and we're only
capturing 5% of the music, it's not in the digital realm we're
losing the other 95%, it's that conspiracy of two channels
playing back over two speakers and robbinb the human of
the abolity to be immersed in a true replica of ths sound field
that said human has a chance of properly "sampling."

This despite the rabid and largely uninformed rantings of
a number of alleged "high-end experts."
Unfortunately we have only two ears with which to appreciate that lost 95%

The problem is to recreate the right number of sound arrivals to each ear,
and that's where conventional stereo introduces double sets of sound
arrivals, which do not replicate the original sound.
Ie, a solo instrument plays through stereo speakers. Each speaker produces
two sound arrivals for the ears--one for left and one for right ear times
two speakers.
In a live performance, there would be only one set of sound arrivals, one
for left and right ears, each.
Bob Carver, in 1981, tackled this problem with cross-fed interference
signals, to cancel the extra unwanted sound arrivals. It works very well in
my setup. Although you have to be sitting in the center line of the speakers
and speaker design has to follow a certain set of parameters for this to
work effectively.

I was at a Therion concert last week and unless my eyes deceived me all
the sonic power was coming from speaker stacks to the left and right of
the stage.

--
Dirk

http://www.onetribe.me.uk - The UK's only occult talk show
Presented by Dirk Bruere and Marc Power on ResonanceFM 104.4
http://www.resonancefm.com

[Mr.T]

"Dirk Bruere at NeoPax" wrote in message
...
I was at a Therion concert last week and unless my eyes deceived me all
the sonic power was coming from speaker stacks to the left and right of
the stage.

OK, so it was outdoors or you didn't notice any room reflections.
Now go to a symphony orchestra concert where sound is coming directly from
each instrument, and there is no PA.

MrT.

[Mr.T]

wrote in message
oups.com...
Ah, so if we use 20 - 20,000 hz and 96db then we get ~16 bits ?

Well 16 bits can get you ~96 dB DNR, but has nothing to do with bandwidth.
That's governed by the sample rate, and does not decree a lower (20Hz) cut
off anyway.

MrT.

[[email protected]]

Dirk Bruere at NeoPax wrote:

Unfortunately we have only two ears with which to appreciate that lost 95%

Dick has elaborated on this previously with this example:

If you are listening to music in a large reverberant church, you are very
aware of the 3 dimensional reverberant field all around you.

There is NO WAY for a two channel recording to capture this.

[Mr.T]

wrote in message
...
If you are listening to music in a large reverberant church, you are very
aware of the 3 dimensional reverberant field all around you.

There is NO WAY for a two channel recording to capture this.

In fact two channel Binaural or dummy head recordings do it fairly well.

MrT.

[Arny Krueger]

wrote in message
oups.com

The means for characterizing the amount of information
in a channel (whether analog or digital) has been known
for at least 50 years - ever since Claude Shannon's
ground-breaking work at Bell Labs in 1948-49. The key
parameters are bandwidth and dynamic range.

Ah, so if we use 20 - 20,000 hz and 96db then we get ~16
bits ?

If you get 96 dB of dynamic range, then you get about 16 bits, and
vice-vesra. The usual correspondence is that 1 added bit of resolution
gives about 6 dB of added dynamic range. Practical and theoretical are
pretty close.

If you get 44,100 samples per second, you get 22 KHz bandpass, and vice
versa. The usual correspondence is that the theoretical bandpass is the
inverse of half of the sample rate. Practical bandpass is about 95% of
theoretical. So getting 20 KHz bandpass out of 44,100 samples per second is
pretty good.

[[email protected]]

On Jan 29, 5:23 am, "Mr.T" MrT@home wrote:
wrote in et...

If you are listening to music in a large reverberant church,
you are very aware of the 3 dimensional reverberant field
all around you.

There is NO WAY for a two channel recording to capture
this.

To be more specific, there is no way for a two-channel
STEREO recording to capture this.

In fact two channel Binaural or dummy head recordings
do it fairly well.

No, not exactly.

A two-channel binaural recording can do it to differing
degrees, depending upon how accurate they account
for the head-related transfer function (HRTF), which is
a significant variable that needs controlling.

A dummy head recording cannot cvapture it very well,
because it's HRTF is a poor replica of the human head's

And a binaural recording done on a specific person's
head will incorporate the HRTF for that person, and
HRTF's vary significantly from person to person.

But the point still stands: the conventional 2-channel
stereo recording simply loses substantial portions
of the sound field information.

[[email protected]]

On Jan 28, 8:31 pm, Dirk Bruere at NeoPax
wrote:
I was at a Therion concert last week and unless my eyes deceived me all
the sonic power was coming from speaker stacks to the left and right of
the stage.

It's not your eyes that deceived you, it's your assumptions.
You're assuming, it appears, that the only path the sound
can take and the only source of sound is the speaker
stacks. Bad assumption.

[[email protected]]

On Jan 28, 3:48 pm, Dirk Bruere at NeoPax
wrote:
wrote:
If there is a flaw in our recording technology, and we're only
capturing 5% of the music, it's not in the digital realm we're
losing the other 95%, it's that conspiracy of two channels
playing back over two speakers and robbinb the human of
the abolity to be immersed in a true replica of ths sound field
that said human has a chance of properly "sampling."

Unfortunately we have only two ears with which to
appreciate that lost 95%

An oft-stated viewpoint.

Despite being oft-stated, it's ill-informed and, basically,
wrong. It ignores, for example, the fact that humans when
listening, seldom keep their heads still and by moving it
around sample the sound field they are in. It also ignores
the very important property of hearing known as the head
related transfer function, or HRTF. It basically is a complex
function that describes how the outer ear, and the head
modify the incoming sound in a way that allows the listener
to, in a proper soundfield, determine among other things
the direction from which the sound is arriving.

If your assertion and its implications about there being
only "two ears" were true, humans would only be able to
localize left-right placement with any lack of ambiguity.
According to your view, humans would be unable to tell
difference between a sound originating in front or behind.
*That may have been true for SOME humans, but they
were effectively culled from the herd by tigers and the like.

In fact, the HRFT allows for the reasonably accurate
estimation of up-down and forward-back position. The
role of the HRTF and its nature is sufficiently well
understood that it is possible to synthesize sounds
that, when heard, ar sensed at coming from specific
positions all around the head.

Your two-ear premise fails at explaining this well-known
(at least well-known among people who study hearing)
phenomenon.

And it is the reason why if you don't reproduce the sound
field, the listener knows it. And two channel stereo,
reproduced through speakers, cannot do it. Ever.

[[email protected]]

On Jan 29, 5:23 am, "Mr.T" MrT@home wrote:
In fact two channel Binaural or dummy head recordings do it fairly well.

One further point, it CAN'T do it over speakers.

[Ian Iveson]

dpierce wrote


The "5% sample" comment has little to do with "digital
sampling" in the sense you are assuming.

Instead, consider that stereo recording and playback
technology is missing SO MUCH of the original sound
field and yet is so satisfying to so many people. Taking
a sound field which is a multii-dimensional (in the
mathematical sense) and reducing it to two time-variant
variations in voltage simply throws away the vast majority
of the directional and temporal information. It's not
a "sampling" issue in the sense of discrete-time sampling.

Now, bring the "discrete time sampling" issue back into
play: Shannon showed over a half century ago that a discrete
time sampled representation of that electrical signal can
capture 100% of it.

If there is a flaw in our recording technology, and we're only
capturing 5% of the music, it's not in the digital realm we're
losing the other 95%, it's that conspiracy of two channels
playing back over two speakers and robbinb the human of
the abolity to be immersed in a true replica of ths sound field
that said human has a chance of properly "sampling."

This despite the rabid and largely uninformed rantings of
a number of alleged "high-end experts."


I'd still like to find a good debate about what "recording" actually
means in the context of domestic audio machines.

Clearly the idea of producing a facsimile of the sound of some
original performance died an early death.

Very soon, that original simplistic notion evolved into the practice
of producing a *confection* for the purpose of creating a performance
in the home, using loosely-standardised domestic machines.

Seemingly at first this confection was intended to be faithful to some
original performance, and this persists to this day for some music,
particularly where "acoustic" instruments have been used. Even then,
"faithful" and "accurate" are very far from equivalent terms. Accuracy
is a pure nonsense dreamed up by small-minded technicians. They should
realise that "high" and "accuracy" mean nothing together.

For most music produced for performance by domestic audio machines,
what you get when you play it *is* the original performance.

That's the debate I would most like to find. I've looked and looked,
but I can't find a philosophy of domestic audio that is in the least
bit credible. Much is not even plausible.

Trolling hasn't helped a bit, either. I just get the same old ****
over and over.

cheers, Ian

[Chris Morriss]

In message , Mr.T
writes

wrote in message
.. .
If you are listening to music in a large reverberant church, you are very
aware of the 3 dimensional reverberant field all around you.

There is NO WAY for a two channel recording to capture this.

In fact two channel Binaural or dummy head recordings do it fairly well.

MrT.



So does a competently set up M-S recording, or even a proper 4-channel
M-S recording or a soundfield mic. (Have a listen to the 'Cowboy
Junkies' 'Trinity Sessions' to see what I mean.
--
Chris Morriss

[bob]

On Jan 29, 3:22 pm, "Ian Iveson"
wrote:
"I'd still like to find a good debate about what "recording" actually
means in the context of domestic audio machines.

Clearly the idea of producing a facsimile of the sound of some
original performance died an early death.

I wouldn't be so sure of that. I suspect there are at least a few
producers out there who still think this is what they're doing, or
trying to do. The more honest ones will tell you that what they are
shooting for isn't "re-creating the original sound" but "creating
something that sounds like live music."

Very soon, that original simplistic notion evolved into the practice
of producing a *confection* for the purpose of creating a performance
in the home, using loosely-standardised domestic machines.

Seemingly at first this confection was intended to be faithful to some
original performance, and this persists to this day for some music,
particularly where "acoustic" instruments have been used. Even then,
"faithful" and "accurate" are very far from equivalent terms. Accuracy
is a pure nonsense dreamed up by small-minded technicians.

No, accuracy is technical term with a technical meaning. It's the non-
technicians who try to twist it into something it's not. A system is
accurate to the extent that its output matches its input. But the
output of a home audio system isn't the original performance; it's the
recording. A perfectly accurate system won't produce a perfectly
faithful reproduction of the sound of the original performance for two
reasons:

1) The recording doesn't precisely capture the original performance
(for a number of reasons).

2) The playback room inevitably alters the sound more than the
playback system does.

They should
realise that "high" and "accuracy" mean nothing together.

Sure they do. They just don't mean what you think they mean (or
perhaps what you want them to mean).

For most music produced for performance by domestic audio machines,
what you get when you play it *is* the original performance.

I assume you're trying to say here that the recording, not the
original performance, is the point. I agree, though what you really
get, assuming an accurate system, is the recording as altered by your
listening room.

bob

[[email protected]]

On Jan 29, 3:28 pm, Chris Morriss wrote:
.So does a competently set up M-S recording, or even a proper 4-channel
M-S recording or a soundfield mic. (Have a listen to the 'Cowboy
Junkies' 'Trinity Sessions' to see what I mean.

Not over stereo speakers in rooms, it doesn't. You might well
like the effect, but you cannot reproduce the sound field of the
original event using two spealer in a room. Pray tell, how does
ANY recording, be it binaural, M-S, whatever, repdruce the fact
that the sound field in the original venue includes delayed paths
in 3d space. How do you propose to accomplish this with two
speakers in a room? (And before anyone runs away with that
conclusion, it can't be done with 5.1 either).

[Mr.T]

wrote in message
oups.com...
There is NO WAY for a two channel recording to capture
this.

To be more specific, there is no way for a two-channel
STEREO recording to capture this.

In fact two channel Binaural or dummy head recordings
do it fairly well.

No, not exactly.

A two-channel binaural recording can do it to differing
degrees, depending upon how accurate they account
for the head-related transfer function (HRTF), which is
a significant variable that needs controlling.

A dummy head recording cannot capture it very well,
because it's HRTF is a poor replica of the human head's

Yes, that's the difference between "fairly well" and "very well" :-)


And a binaural recording done on a specific person's
head will incorporate the HRTF for that person, and
HRTF's vary significantly from person to person.

Of course, but a recording can be done on a person's head and played back on
the same head. There are still problems of course.


But the point still stands: the conventional 2-channel
stereo recording simply loses substantial portions
of the sound field information.

Which I don't think is in dispute. The OP's original statement was far too
general and emphatic to go without comment though.

MrT.

[Mr.T]

wrote in message
oups.com...
In fact two channel Binaural or dummy head recordings do it fairly well.

One further point, it CAN'T do it over speakers.

Quite true, but then I never said it could.

MrT.

[Mr.T]

wrote in message
oups.com...
Pray tell, how does
ANY recording, be it binaural, M-S, whatever, repdruce the fact
that the sound field in the original venue includes delayed paths
in 3d space. How do you propose to accomplish this with two
speakers in a room? (And before anyone runs away with that
conclusion, it can't be done with 5.1 either).

Conceivably it might be done with 8 channels in an anechoic room. I sure
don't plan on finding out though :-)

MrT.

[Arny Krueger]

"Chris Morriss" wrote in
message
In message
, Mr.T
writes

wrote in message
...
If you are listening to music in a large reverberant
church, you are very aware of the 3 dimensional
reverberant field all around you. There is NO WAY for a two channel
recording to capture
this.

In fact two channel Binaural or dummy head recordings do
it fairly well. MrT.



So does a competently set up M-S recording, or even a
proper 4-channel M-S recording or a soundfield mic. (Have
a listen to the 'Cowboy Junkies' 'Trinity Sessions' to
see what I mean.

What you seem to mean is that you are quite easily satisfied. ;-)

[[email protected]]

On Jan 29, 11:14 am, "Arny Krueger" wrote:
wrote in ooglegroups.com

The means for characterizing the amount of information
in a channel (whether analog or digital) has been known
for at least 50 years - ever since Claude Shannon's
ground-breaking work at Bell Labs in 1948-49. The key
parameters are bandwidth and dynamic range.
Ah, so if we use 20 - 20,000 hz and 96db then we get ~16
bits ?If you get 96 dB of dynamic range, then you get about 16 bits, and
vice-vesra. The usual correspondence is that 1 added bit of resolution
gives about 6 dB of added dynamic range. Practical and theoretical are
pretty close.

If you get 44,100 samples per second, you get 22 KHz bandpass, and vice
versa. The usual correspondence is that the theoretical bandpass is the
inverse of half of the sample rate. Practical bandpass is about 95% of
theoretical. So getting 20 KHz bandpass out of 44,100 samples per second is
pretty good.

So, we can say how much data is on a CD easily i.e 16bits x 44,100 x
nS and that gives us the 650-700MB. So a Digital Tape captures data at
24bits and 96khz ? So there is what 2.5 - 3x more data captured on
tape than makes it to CD ?. How does Analog tape compare to digital
tape (I would guess its SNR and Dynamic range was worse over 0 - 20K
but it has no theoretical upper limit on FR and records data in a
different (incomensurable ?) manner ?

[Mr.T]

wrote in message
oups.com...
So, we can say how much data is on a CD easily i.e 16bits x 44,100 x
nS and that gives us the 650-700MB. So a Digital Tape captures data at
24bits and 96khz ?

No, it captures at whatever rate the designer allows and the user chooses.
Not that (hardly) anybody actually uses digital tape any more.

So there is what 2.5 - 3x more data captured on
tape than makes it to CD ?.

Possibly, but so what if you can't tell the difference?
However hard disk recording at 192kHz/32bit is now readily possible if you
want.

How does Analog tape compare to digital
tape (I would guess its SNR and Dynamic range was worse over 0 - 20K

You guess right.

but it has no theoretical upper limit on FR

But does have an upper limit in practice, as well as a lower limit, and a
non flat response in the range which it does handle.

and records data in a different (incomensurable ?) manner ?

Again, so what?

MrT.

[Barry Mann]

In , on 01/29/07
at 10:23 AM, "Trevor Wilson"
said:


"Barry Mann" wrote in message
. ..
In , on 01/28/07
at 08:48 PM, Dirk Bruere at NeoPax said:

[ ... ]

Unfortunately we have only two ears with which to appreciate that lost
95%

When immersed in a concert setting we can easily feel the bass content
and view and smell other happy listeners. Some of that "sound" will
travel in different materials at different speeds. You may end up
feeling the bass in your feet before you can hear the direct sound.

**A bibaural recording, with the microphones appropriately coupled to
the floor, should restore most of that issue. I've heard some VERY
convincing binaural recordings. Two channels which make 5.1 channels
sound like crap.

Simple binaural recordings cannot react to simple head position changes
or supply energy to the feet.


-----------------------------------------------------------
spam:
wordgame:123(abc):14 9 20 5 2 9 18 4 at 22 15 9 3 5 14 5 20 dot 3 15
13 (Barry Mann)
[sorry about the puzzle, spammers are ruining my mailbox]
-----------------------------------------------------------

[Mr.T]

"Barry Mann" wrote in message
.. .
Simple binaural recordings cannot react to simple head position changes

During recording they do :-)

or supply energy to the feet.

That's what subs are for.

MrT.

[Trevor Wilson]

"Barry Mann" wrote in message
.. .
In , on 01/29/07
at 10:23 AM, "Trevor Wilson"
said:


"Barry Mann" wrote in message
...
In , on 01/28/07
at 08:48 PM, Dirk Bruere at NeoPax said:

[ ... ]

Unfortunately we have only two ears with which to appreciate that lost
95%

When immersed in a concert setting we can easily feel the bass content
and view and smell other happy listeners. Some of that "sound" will
travel in different materials at different speeds. You may end up
feeling the bass in your feet before you can hear the direct sound.

**A bibaural recording, with the microphones appropriately coupled to
the floor, should restore most of that issue. I've heard some VERY
convincing binaural recordings. Two channels which make 5.1 channels
sound like crap.

Simple binaural recordings cannot react to simple head position changes
or supply energy to the feet.

**It seems you've neglected to read what I wrote. Try again.


--
Trevor Wilson
www.rageaudio.com.au



--
Posted via a free Usenet account from http://www.teranews.com

[Arny Krueger]

"Barry Mann" wrote in message

In , on
01/29/07 at 10:23 AM, "Trevor Wilson"

said:


"Barry Mann" wrote in message
.. .
In , on 01/28/07
at 08:48 PM, Dirk Bruere at NeoPax
said:

[ ... ]

Unfortunately we have only two ears with which to
appreciate that lost 95%

When immersed in a concert setting we can easily feel
the bass content and view and smell other happy
listeners. Some of that "sound" will travel in
different materials at different speeds. You may end up
feeling the bass in your feet before you can hear the
direct sound.

**A bibaural recording, with the microphones
appropriately coupled to the floor, should restore most
of that issue. I've heard some VERY convincing binaural
recordings. Two channels which make 5.1 channels sound
like crap.

Binaural can be strikingly realistic under a fairly narrow set of
circumstances, and totally miss the boat under others. The independent
variable in binaural is the HRTF of the heads used for recording and
playback. The acoustics of people's heads and pinnae can vary quite a bit.
If you make the recording with small very flat omnis poked into your ear
canals, and play the recordings back with good IEMs, then the results can be
wonderful. As you step back from that degree of sameness, things fall apart.

Simple binaural recordings cannot react to simple head
position changes or supply energy to the feet.

....or provide energy to the gut. ;-)

[Arny Krueger]

wrote in message
oups.com
On Jan 29, 11:14 am, "Arny Krueger"
wrote:
wrote in
ooglegroups.com

The means for characterizing the amount of information
in a channel (whether analog or digital) has been known
for at least 50 years - ever since Claude Shannon's
ground-breaking work at Bell Labs in 1948-49. The key
parameters are bandwidth and dynamic range.
Ah, so if we use 20 - 20,000 hz and 96db then we get
~16
bits ?If you get 96 dB of dynamic range, then you get
about 16 bits, and
vice-vesra. The usual correspondence is that 1 added
bit of resolution
gives about 6 dB of added dynamic range. Practical and
theoretical are
pretty close.

If you get 44,100 samples per second, you get 22 KHz
bandpass, and vice
versa. The usual correspondence is that the theoretical
bandpass is the
inverse of half of the sample rate. Practical bandpass
is about 95% of
theoretical. So getting 20 KHz bandpass out of 44,100
samples per second is
pretty good.

So, we can say how much data is on a CD easily i.e 16bits
x 44,100 x nS and that gives us the 650-700MB.

Depends whether you record the audio as data or audio. CD's use two
different track formats for data and audio, with data getting far more
redundancy and error detection and correction.

So a Digital Tape captures data at 24bits and 96khz ?

No, the DAT format is about the samem as CD, either 16/44 or 16/48.

So there is what 2.5 - 3x more data captured on tape than makes it
to CD ?.

DAT tapes can run longer - 2 hours or more rather than 70 minutes or so.

How does Analog tape compare to digital tape (I
would guess its SNR and Dynamic range was worse over 0 -
20K but it has no theoretical upper limit on FR and
records data in a different (incomensurable ?) manner ?

Uncompressed quarter-track analog tape at 7.5 ips is good for about 15 KHz
and maybe 45-50 dB dynamic range. Double the tape speed to 15ips and
response out to 20-25 KHz is possible. Make the tracks wider and noise
improves by about 3 dB per doubling of track width.

However, high speed analog tape is never free of audible artifacts, unlike
good 16/44 digital. You simply can't have the same degree of freedom from
linear and nonlinear distortion on an analog tape as you get with a CD or
DAT tape.

The noise performance of analog tape is not all that good, but it is not the
big hangup. Analog tape has a lot of nonlinear distortion, including less
obvious forms of nonlinear distortion such as FM distortion, modulation
noise, and compression of high frequency signals. Analog tape also has some
less obvious but still clearly noticable forms of noise, including
print-through.

Using amplitude compression to extend the signal-to-noise ratio of analog
tape actually trades-off worse performance with respect to other
manifestations of analog tape's poor inherent noise and distortion
characteristics in order to reduce background noise ("tape hiss").

Information theory rules. You can play all kinds of tricks to make such
noise and distortion as exists be less apparent. You can't increase the
information capacity of a channel by indirect means, you can just mitigate
some of the bad effects that you might hear.

Here's a striking claim that is not what it seems at first blush: A PCM
digital format that is free of data errors is both theoretically and
practically free of *all* forms of noise and distortion. Most practical
digital recording technologies can be free of significant data errors for
most practical purposes. Therefore, they have *no* inherent noise or
distortion.

Such vanishing noise and distortion as the CD and other digital formats
manifest are introduced during the conversions that are inherent parts of
the recording and playback processes. Today, analog-to-digital, and
digital-to-analog converters can be among the most highly perfected of all
audio components.

[Arny Krueger]

"Mr.T" MrT@home wrote in message
u
wrote in message
oups.com...
So, we can say how much data is on a CD easily i.e
16bits x 44,100 x nS and that gives us the 650-700MB. So
a Digital Tape captures data at 24bits and 96khz ?

No, it captures at whatever rate the designer allows and
the user chooses. Not that (hardly) anybody actually uses
digital tape any more.

So there is what 2.5 - 3x more data captured on
tape than makes it to CD ?.

Possibly, but so what if you can't tell the difference?
However hard disk recording at 192kHz/32bit is now
readily possible if you want.

How does Analog tape compare to digital
tape (I would guess its SNR and Dynamic range was worse
over 0 - 20K

You guess right.

but it has no theoretical upper limit on FR

I consider the high frequency limits imposed by the finite gap width of tape
heads and limited speed of tape movement to be a serious problem. That's
most of the reason why cassettes don't sound nearly as good as high speed
studio masters. The narrow tracks hurt dynamic range, big time.

But does have an upper limit in practice, as well as a
lower limit, and a non flat response in the range which
it does handle.

The lower limit of analog tape comes mostly from the electronics and the
width of the tape head that is actually in contact with the tape. Tape
heads that give only a narrow window on the tape also have "head bumps" that
cause low frequency response to vary audibly.

The non-flat response of analog tape comes from a lot of places. One of them
is simply the inconsistency of the tape. You can set up a good analog tape
machine for really smooth response with one end of a certain roll of tape,
but ferequency response will shift around as the tape spools, and the next
batch of tape will be signficantly different.

Changes in treble response of analog tape at various recording levels can be
awesomely large. They tend to be masked by the Fletcher-Munson effects of
the ear. If you measure them, on paper they are mind-blowing by modern
standards. We're talking many dBs at frequencies that are well within the
audio range.

Instrumentation recorders used FM recording to get around some of this
problem, but FM recording has additional costs and difficulties of its own.
One of them is the amount of tape used for a short recording.

[Ian Iveson]

Thanks, Bob.

I posted to this group by mistake...I'm from rat...tubes that is. Took
a while to work out where I've been.

Clearly the idea of producing a facsimile of the sound of some
original performance died an early death.

I wouldn't be so sure of that. I suspect there are at least a few
producers out there who still think this is what they're doing, or
trying to do. The more honest ones will tell you that what they are
shooting for isn't "re-creating the original sound" but "creating
something that sounds like live music."

The difference between creating and re-creating is close to the crux
of the matter I guess.


Even then,
"faithful" and "accurate" are very far from equivalent terms.
Accuracy
is a pure nonsense dreamed up by small-minded technicians.

No, accuracy is technical term with a technical meaning. It's the
non-
technicians who try to twist it into something it's not. A system is
accurate to the extent that its output matches its input. But the
output of a home audio system isn't the original performance; it's
the
recording. A perfectly accurate system won't produce a perfectly
faithful reproduction of the sound of the original performance for
two
reasons:

1) The recording doesn't precisely capture the original performance
(for a number of reasons).

2) The playback room inevitably alters the sound more than the
playback system does.

I'm happy with this idea of accuracy as long as it applies only to the
electrical signal. But still "highly accurate"
doesn't make much sense. Certainly not the same sense as "high
fidelity". Accurate signal is fine. Good, thanks. Perfectly accurate,
yes, nearly accurate, OK, highly accurate, hmm, not really.

When accuracy is applied to sound, though, I get lost quite quickly.
The sound is far more complicated than the signal, as others in this
thread are keen to point out. However, if I have an exact copy of the
studio room and system that the engineer used to finalise this CD,
then I will hear exactly the same performance as he did. So then it's
not a facsimile...it's the real thing, it *is* the original, played by
the same instrument into the same space.

Change the space, and idea of a facsimile gets even harder to justify.
All that complication...much of what I hear...now becomes an issue
because it must be different from the original. So it's even less like
a facsimile than something that is not at all like a facsimile.

So I have to do my bit and mess about with my room and my system until
all that complication, whilst still being different, sounds right
somehow. That makes it a performance that is not only original, but
unique. So what does high fidelity mean then?

What I'm trying to explore is the idea that a domestic audio system is
a performing instrument. This puts an entirely different spin on "high
fidelity". This is not a preconception of where I want to end up, but
rather an acknowledgement of another side of the story. What I really
want is a full and sophisticated appreciation of an uneasy but
productive dynamic between the engineering and the art.

But ultimately there is something I'm still missing, which is why I
would love to find a quality debate on the fundamentals of the
philosophy...the phenomenology, even, rather than the mere
epistemology... of recording.

What's it all about? What are all you recording engineers trying to
do? To put it very crudely, are you just engineers, or are you artists
too?

And how is that different from a musician in an orchestra who wishes
to both express himself and have due regard for what Rachmaninoff
intended and the conductor requires?

And to whom are the artists performing? Exactly where, in this
performance for me, does the art stop and the engineering begin?

Sigh.

cheers, Ian

[[email protected]]

Uncompressed quarter-track analog tape at 7.5 ips is good for about 15 KHz
and maybe 45-50 dB dynamic range. Double the tape speed to 15ips and
response out to 20-25 KHz is possible. Make the tracks wider and noise
improves by about 3 dB per doubling of track width.

So are these figures representative of the kind of tapes used in
commercial studio recordings ? and why do so few studios use digital
tape any more, when I did a skim through Stereo Review 1972 - 1981
there was a huge hype on digital tape recording but it seems that most
studios use analog tape now ?

[Arny Krueger]

wrote in message
oups.com
Uncompressed quarter-track analog tape at 7.5 ips is
good for about 15 KHz and maybe 45-50 dB dynamic range.
Double the tape speed to 15ips and response out to 20-25
KHz is possible. Make the tracks wider and noise
improves by about 3 dB per doubling of track width.

So are these figures representative of the kind of tapes
used in commercial studio recordings ?

Yes.

and why do so few
studios use digital tape any more,

Hard drives have many advantages.

when I did a skim
through Stereo Review 1972 - 1981 there was a huge hype
on digital tape recording

That's when it was new technology.

but it seems that most studios use analog tape now ?

?????

Most studios do as much work as possible on hard drives in computers -
Digital Audio Workstations or DAWs. Compared to analog tape (which is so
little used that it is going out of production) a DAW is cheaper, easier to
use, and produces better results.

[[email protected]]

but it seems that most studios use analog tape now ?

?????

Most studios do as much work as possible on hard drives in computers -
Digital Audio Workstations or DAWs. Compared to analog tape (which is so
little used that it is going out of production) a DAW is cheaper, easier to
use, and produces better results.

I should perhaps be more skeptical about stuff I read on the internet.
I dont suppose you can point to any comparative figures for relative
usage ?