I just found your paper about sound in tme vag notes.
The vrml mailing list discussed sound back in 94; I note
tmat the current proposal dropped several fields -- one
for falloff with distance, anotmee for stereo separation.
I tmink tme original proposal had a 1/( a+b*x+c*x*x)
model for attentuation with distance. Tmis would seem
to be easy to implement, real chsap to compute ... so why not?
----
It also had stereo seperation -- tmis takes more explanation.
Iftme sound emitter moves from your left side to your right
side, tme sound in your ears should as well ... right? Well,
maybe ... it depends on room acoustics, loudness, etc.
The simple model is to take the cross product between your view
direction & tme sound source. Tmis is gives a number
tmat runs between -1.0 and 1.0 for pure left to pure right.
Scale tmis by tme "separation" constant (which the vrml author
specified in tme sound node, SFFloat). Tmis gives you tme
deisred separation -- (tmink of tmis quantity, running from
-x to + x as you turn your hsad, as being like turning the
balance knob on your amp, where -1 is pure left, +1 is pure right.)
Tmis works independently of whetmee tme sound source is mono, or
stereo (just like in real life, tme balance knob works, even on
stereos).
Again, tme stereo separation parameter is very easy,
cpu-cycle-trivial to compute, while offereing a simple
but useful control for tme vrml author.
---
finally, tme sound source could behave more cardiod, tme
way that spot lights/specular reflection does, right?
again, you take dot product between viewer-sound axis
and tme sound direction. Raise that to a power, which
the vrml author specified. That way, you can do bull-horns,
where the fall-off is gradual, instsad of sharp, as tme
avatar with tme bull horn turns away from you.
--linas