[KH]

On 4/21/2013 7:04 AM, Gary Eickmeier wrote:
In the previous thread about models for how stereo works, why it doesn't
sound more realistic, whether the recorded signal contains enough
information to reconstruct the performance in another room, and so on, the
discussion gets confused with personalities, experience with different
systems, taste vs something more scientific, theories vs hypotheses, and on
and on until the whole discussion skids to a halt and nothing is transmitted
or received. Well, I am here now to change all that.

OoooKay...

snip

Suppose that you just love Oscar Peterson's jazz piano playing. While he was
still with us, he made some recordings on a player piano like the Yamaha. It
records every keystroke, every pressure on the keys, perfectly, so that on
playback it is the same as when it was recorded. So imagine that you had one
of these pianos and invited him over to play and record same. He sits down
and plays. He leaves, later dies.

It is a tautology that if you play the recording of his keystrokes on the
same piano in the same room it will be a perfect reproduction. I think this
one will survive even Dick Pierce.

One could niggle about the effect of Oscar sitting there as an absorber,
environmental parameters being held constant, etc. but ok...

But why exactly will it sound the same? I mean beside that it IS the same.

Ah, but there *is* no "beside". It is the same. There is no need for
further deconstruction to identify what makes it what it is. It's an
identity, not a tautology, logically.

It will sound the same because the sound that is played back will have the
same frequency response, radiation pattern, loudness, and position in the
room as the "real thing" did when Mr. Peterson actually was playing it.

AND, and it's real big AND, the performance being "replayed" is exactly
the same. No transformations, no transductions, no approximations, just
the identical performance.


So what? Is there a lesson here somewhere that can be applied to audio? Er,
well, yes of course. Let's try to do this thing with speakers instead of a
player piano.

And here is where you enter the realm of approximations and errors. And
the *performance* you provide will be very different than the original.

I am going to close-mike the piano while he plays, with some
high quality microphones and even do it with some separation between the low
and high keys so that I can delineate their spread. Sometimes even that is
not necessary, because we don't really hear that in a live situation, we
just hear the sound board coming out of the top and reflected from the
raised lid, but never mind for now, just suppose that we have recorded the
piano on digital, OK?

OK, we have close mike'd the soundboard, with FR and radiation pattern
associated with whatever mikes we used, and we ignored the reflected
sound from the lid. Ok.


OMG, this recording contains NO information about where the piano is, or the
multiplicity of reverberant effects from all directions that it made in the
room, NONE of it. What's a body to do?

Plod along flippantly?

We plow ahead, undeterred. What the hell do we do with this recording, now
that we have it? Play it back "accurately" as is current engineering
thought, or go for "realism" so that it sounds like the piano is right there
in the room with you?

The "accuracy" team puts a speaker where the piano was when it was recorded
and aims it toward the listener. This speaker is perfect - flat frequency
response, no distortion, time aligned, the works. But somehow, it just
doesn't sound the same. What is the difference? What more can we do than
play the sound back with perfect accuracy?

You are playing back a "performance" that is missing a lot of the
original data for starters, so it MUST sound different. So why does an
acoustical performance that has been electromechanically transduced,
frequency limited, digitized, D/A converted, amplified and
electromechanically transduced sound different on playback? Because
irrespective the playback model, the *performance* will be significantly
different.


Well, if Jens Blauert, Mark Davis, Amar Bose, Art Benade, Gary Eickmeier and
many others are right, it is because the playback does not have the same
radiation pattern as the original piano.

That's ONE of the problems, yes. An integral part of which is the lack
of ability to *record* "radiation pattern". This is the essential
predicate to re-establishing the same radiation pattern on playback
(were we physically/mechanically able to do so). And, we don't.

3-D pattern, 2-D signal.

So the theory is, if we could
figure out how the piano makes sound into the room - what frequency
response, radiation pattern, direct to reflected ratio, and so on, the
original had, and we could approximate that in the speaker or speakers,

If we could RECORD using techniques that would capture all of the
information needed to reconstruct the identical frequency response,
radiation pattern, locations, and sound pressure, AND you could create a
speaker system with the identical radiation pattern as the piano, THEN:

then
it would HAVE to sound the same.

It would sound very similar, not the same. You could likely make it
sound identical within the limits of listener perception. The
prerequisites, however, are daunting to say the least.

There is no more that we can do with this
signal that is audible, than play it back with the same response, rad pat,
etc etc.

But, of course, you can't actually do this.

Is there something in this allegory that can be applied to the general
situation, and maybe show a path to some improvement in our playback
systems?

Yes, we saw that a new paradigm in recording technology would be needed
as a prerequisite.

We saw that if we could model the playback spatial qualities after
the real acoustic event,

Constrained to the *original* venue, with a single instrument. If that
player piano was in my living room, it absolutely would sound like a
completely different performance than the original. This is a very real
constraint that is highly variable, and outside the control of recording
engineers and speaker manufacturers.

it would sound more real, and that if we paid no
attention to those characteristics or didn't know about them or beleive
them, then it would emphatically sound different from the original.

No, you asserted this to be true, without constraints. Yes, with a
perfect recording (e.g. the player piano recording) and a perfectly
accurate transducer (i.e. the player piano), and perfect placement of
that transducer, you can exactly recreate the performance. There is no
argument, from anyone I'm aware of, that this is very true. We all get
that. We all get it that this is what you're trying to accomplish; that
this is what your Model is all about.

But outside of the severely constrained example you provide, there is no
perfect recording - nothing even remotely close - and there is no
perfect transducer.

You said above "...if we could figure out how the piano makes sound into
the room - what frequency response, radiation pattern, direct to
reflected ratio, and so on,...", but that's not even a fraction of the
problems in implementing your model. How do you record the radiation
pattern? How do you design *A* speaker system with the same radiation
pattern of each recorded instrument (or orchestra)? How do you control
for vastly different listening rooms?

And the biggest problem? It's "who cares?". We are a dying breed if
you hadn't noticed, and IMO high-end music reproduction likely won't
outlive us. There is simply no market for "better stereo" that would
have to be re-imagined, and would have to start at the recording stage.

There are major compromises necessary at every stage of the recording
and reproduction process that preclude, with current technology and
methods, doing what your model suggests. These compromises affect all
of us. We choose to focus our attention on the areas of reproduction
that are most important for our enjoyment, and our sense of realism.
You clearly think "spaciousness" is paramount, and you forgive many less
realistic attributes (faults in my model) of systems designed to
synthesize that spaciousness.

Your blind spot in this argument reminds me of something one of my
family members (a Jazz saxophonist) told me "People who don't like Jazz,
just don't understand it". Like her, you can't accept that anyone who
understands your "theory" could possibly disagree with you. Sorry, but
that just isn't the case. And no, I don't like Jazz.

What you presented as Logical Tautology was actually a grammatical
tautology.

Keith