[KH]

On 11/9/2013 9:17 AM, news wrote:
"KH" wrote in message
...
On 11/8/2013 8:07 AM, Gary Eickmeier wrote:

snip

Sorry, I don't want to trash all of your work on the X, A, B, C diagram, but
I'm getting Deja Vu all over again here.

Perhaps if you tried to understand the ramifications of the illustration
instead just dismissing it because it disagrees with your "notion" of
what happens, it might be more fruitful.

I said that all of the foregoing just scratches the surface, so permit me to
elaborate just one more level.

There are two stereophonic recordings being made at the live event. One, the
direct sounds and their directions, left to right. Two, the reflected sounds
have been recorded in stereo as well - as you have agreed.

No, I have NOT agreed with this. I have stated that there is one, and
only one, stereo recording being made, and that recording contains both
the direct and reflected sounds superimposed upon one another, the sum
of which is reflective of their relative delay times.

Sure, assuming that there are *two* separate stereo recordings is quite
useful for your model, but is simply not factual.

Both the
directions and the timings of those reflections, plus the reverberant tail
as the sound dies out during the reverb time of the hall.

Irrelevant to the discussion as far as I can see.

No, we cannot find enough channels to do each reflection with a direct
loudspeaker, nor could we do a new setup for every recording. What is
important to audibility is that these reflected sounds NOT come from the
same directions (sources) as the direct sounds that were recorded.

OK, here we agree. Now, ponder this a bit further in the context of the
illustration I provided. Given that the reflected sounds are
superimposed on the direct sound in the original recording, the direct
portion of the sound in the playback room WILL have those reflected
components coming DIRECTLY at you as part of the direct, first arrival
soundfield. Yet you state this is unacceptable, and doesn't happen. But
you can provide no physical mechanism that accounts for this.

You simply cannot continue to pretend that this is not the case unless
you provide some evidence for how these reflected sounds are
disambiguated from the direct sound. It is physically impossible.

snip

The main point is that the reflections from the recording come from correct
spatial angles on playback, which MUST be different from the direct sounds.
Your whole timing discussion is mostly irrelevant to this spatial
requirement.

And all of those reflections will contain BOTH the reflected and direct
sounds from the original recording.

The fact that the timings of the reflections in your room are
shorter than those in the recording matters not. Those timings remain as
recorded, but now come from spatially similar angles during playback if
properly recorded.

You state this over and over, and ignore simple evidence that this is
simply false.

At this point you declare that Eickmeier is full of it once again, because
there is no way to separate out the reflections from the direct sound in the
recording in order to make this fantastical model work. You would be wrong.

Then provide some explanation supported by physics.

It is all done with time delay. Much like a Hafler or Dolby Stereo playback
system can extract the ambience from a recording and wrap it around to the
sides and rear, the time delay of the reflected sound in your room can
decode the reflected in the recording for the frontal soundstage. How? There
can be only one first arrival, and that HAS to be from the actual speakers
in the model.

And that frontal soundstage that you construct will comprise both the
direct and reflected sound as recorded in the venue.

Later arriving, longer duration sounds will come from the
virtual sources or reflections in your listening room if the balances are
all about right.

You can *create* new soundfields with your approach, and you may like
the outcome. You simply are NOT "extracting" or "decoding" reverberant
information from the original recording.

You have one amalgamated signal comprising both direct and reflected
sound. You cannot selectively add/remove delay times to retrieve the
reflected sound embedded in the recording. Any delay times you build in
the "spacial model" will apply equally to both the direct and reflected
components *of the recorded signal*.

The audible result is that for tight dry recordings with none or little
reverberance contained in them, the direct sounds just take on a harmless
image shift

"Just" a harmless image shift?

toward the reflecting surfaces and form themselves as an aerial
image slightly behind the plane of the speakers. More reverberant recordings
take on the spaciousness and depth

No one denies that your approach can create a sense of spaciousness
(typically way to large IME). It brings other bugaboos with it however,
like those "harmless image shifts" that others find very objectionable.

similar to the live situation and for
similar reasons.

This is simply your opinion. Others don't find the level of "similarity"
that you do.

snip
Probably a few gaps in all that

I'd say.

Unless you can provide a physical mechanism (not an analogy or mental
image) for how the reflected and direct sounds are disambiguated upon
replay - and I'm not talking about how you *create* NEW
reflected/delayed soundfields - then there's really nothing left for us
to discuss. Both the direct and reflect sounds in your model will
contain *both* the direct and reflected sounds recorded in the actual
venue. You like that, I don't.

Keith