[ This text is a revision (August 17th 1993) of the article i posted:
 Date: Mon, 25 Jan 93 1:03:24 EST
 From: dionf@ERE.UMontreal.CA (Francois Dion)
 Message-Id: <9301250603.AA26790@brise.ERE.UMontreal.CA>
 Subject: Franky in 3D
 To: Ultrasound Daily Digest 

 I changed it to reflect the new informations i got. Archer did send me this
 time a nice informational package.
]

Well, i think it's time i put my grain of salt on the 3d mania.

First of all, even if it is a bit long, i suggest to read it all. And if you
dont want to wait for a 3D driver for the GUS, dont forget to read one of my
other post in this digest with the subject: Franky in surround. Also found
under the filename surround.txt. [Note: 3D drivers are now available for
programmers]

Now, on with the show:
There are several systems that are in use to get 3D sounds on recordings and
some have been around since the 50s. Now i wont go into the "how it works"
of the more recent ones, but i think this will clear up some confusion. The
first part is a "hands-on" experiment, the second is informations, including
the address and phone of the owner of the technology that is used with the
Gravis Ultrasound.

Let's get back to the early days of stereo. One record company (i cant seem
to remember) was pushing it, while another (again, blank. anyone?) competed.
Interestingly enough, technological development was put on stereo, and not on
the first 3D system which was called "binaural recording" and it simply
consisted of two microphones placed like the ears. You can try it this way:

Go to a hat store and buy an extruded foam mannequin head. You'll then need
two microphones. Condenser will do, but you will need to power them if you
want to use them with the GUS, since it take a dynamic microphone because
it does not supply phantom power like some mixer with XLR plugs. I will post
a circuit later for Radio-shack condenser mike unit (a small element that
cost about 2$) if there is some interest. (this circuit can be found in the
file micro.arj on archive.epas.utoronto.ca. If you dont want to mess with
that, go with a cardiod dynamic element. Note that sensitive enough cardiod
will cost you a lot, so think about that. You cut holes in the ears of the
head, to insert the microphone units (dont forget to make the wires of the
elements go inside the head and out the rear (or wherever). Use glue to fill
the crack around the mic. Also, the more the ears look like real ears, the
better it will work. If you trim the foam, dont forget to use an hairdryer
to soften it (it will be more uniform). That's it. Try recording sounds, and
you'll be surprised. I was! I did the experiment with a polystyrene head
on which i incrusted two PZM microphones.

Now that you understand how 3D recording is nothing like stereo recording,
we'll see what is accesible presently. 

First, the gadget we just built in the previous section exist commercially,
and is called "Mikey" and is made by Spherical Sound. It's the only system
commercialised where the microphones are placed in a head.

Another system is made by Virtual Audio and claims to enhance stereo depth,
but is not labeled 3D audio. I dont have much more info on it, but from the
description it looks like the same thing as the "mikey".

A few other systems use less restraining microphones situation and can also
be used on any signal because a DSP simulate a 3D signal from parameters
entered on the machine. The DSP will calculate filters based on head related
transfer functions. To get the parameters for these functions, lots of tests
in anechoic chambers had to be done, with frequency, phase and volume
information measured in the listeners ears (actually some kind of "mikey"
is used to get the measures).

QSound (no hyphen) was developped in Alberta, and the inventor sold the
concept to another company (Archer Communications). It will not work at all
with headphones, or if you are not positionned correctly while listening to
it thru speakers and you are not in the soft spot. As for spacial repre-
sentation, it cannot recreate a sound behind the listener, but can enhance
the virtual representation between the listener and the speakers.

And for trivia: Madonna, Sting, Wilson Phillips, Julian Lennon and Paula
Abdul to name a few have used the QSound on their latest recordings.
You cant buy a Qsound system presently but you can rent one and it is now
midi controllable. The Qsound system has been licensed to Creative Labs for
inclusion on their SB16asp cards and has also been licensed to Capcom and
Analog Devices. Strangeley enough, CDROM titles for Sega home entertainment
platform will be released with Qsound, while on this same system you will be
able to get a VR helmet *with headphones* (remember, Qsound doesn't support
that...).

Another variant with a DSP is Roland RSS (Roland Sound Space). It is
better (depending on how it is used) than QSound with headphones, but suffers
the same faith as QSound when you are listening with speakers. Just move a bit
from the soft spot, and suddenly what was in front left is now back left. RSS
was used on Suzanne Cianni _Hotel Luna_ album. RSS cost around 40K$ and is
midi controllable. I have not seen much things done with it yet, so Qsound
and Focal Point seems to be more viable solutions (considering cost!).

Another system on which i have zero information is called Audio Cybernetics.

I also found out that a company called Sixgraph, from Montreal, QC will
release a soundcard for PCs at Comdex in october 93 and from what i was told,
they use their own technology for reproducing 3D audio.

Then you have the convolvotron, a 300 mips 3D engine for PCs. Used a lot in
virtual reality, but can only have 1 3D sound. It is also quite expensive and 
use more than 2 speakers. A less expensive product from the same company
(Crystal river engineering) is the beachtron (around 2000$). They also have
a high end 3D audio workstation, rated at 2.5 gips with a price tag almost
in the 6 digits!

The last technology is called Focal Point 3D Audio. It was developped by Bo
Gehring and first used on the Macintosh computers with a modified
Audiomedia (Digidesign). It cost around 1400$ in this configuration.
Gravis saw that (Focal Point is from Seattle) and it is the system that is
now available on the GUS. At a much better price.

The system produce the sounds with these parameters: direction, elevation
and distance. RSS and Qsound cant accurately represent virtual sound behind
the listener, but FP can, which makes it a better choice for virtual reality.
However, in the midi implementation Gravis will have to develop a SYSEX
command set. We already need it badly, but with 3D, i will shoot myself if i
cant control it thru sysex. [Note: it will probably never be implemented to
the windows driver, unfortunately]. A more complete description of focal
point can be found in the Gravis Ultrasound FAQ. Apparently DOOM will be
the first game to support 3D sound on the Gravis (and probably on PCs).
Sense8 world builder now has Gravis Ultrasound 3D audio support, and
several titles are under work with 3D sound. Several 3D demos are also
available.

Focal Point(tm) 3D audio
1402 Pine av., #127
Niagara Falls, NY 14301
Voice/fax: 1-416-963-9188

I have found most other addresses and once i have a bunch more info, i will
post a totally new article, and hopefully i will have tried all the 3D
systems other than Focal point on computers and Qsound and RSS on CDs (plus
of course binaural recording). In particular, i am curious about the
sixgraph card...

Ciao,
Francois Dion
cism@ere.umontreal.ca
dionf@ere.umontreal.ca
dionf@nyongwa.cam.org
Fido: Francois Dion 1:167/228