> Lots of interesting stuff deleted (I know very little about sound in virtual
> worlds-- one general comment: a lot of the issues you raise are really
> quality of browser issues; smart browsers will try to simulate virtual sound
> in virtual environments more realistically...).
Thinking about sound is excellent, but let's be careful how much, and how
little we state in the specification. Simulating multi-source
doppler-shifted spatialized sound is non-trivial computationally, so we
can't require every browser to do it. On the other hand, if they can do
it, we should be sure to specify the aural effect the author desired.
The situation is akin to specifying a color for an object. The VRML author
specifies the color they want. The browser may be forced to dither down
to a less correct version of the color, but it is not at liberty to
change, say, red to blue.
I think the sound ideas expressed to date on the VRML mailing list are
cool, but let's be careful. I urge interested people to pause and seek the
input of professional sound synthesis people. Please subscribe to the
SIGSOUND mailing list. SIGSOUND info may be found at:
http://www.acm.org/sig_hp/SIGSOUND.html. There are also numerous excellent
books on sound synthesis and sound perception. If people are interested,
I'd be happy to post a bibliography of books and conferences on the subject.
As a technological point of reference, simulating real acoustic
environments, with frequency attenuation, blocking objects, and multiple
moving spatialized sources is currently beyond the abilities of any but
the most expensive dedicated DSP platforms. A Crystal River Engineering
system, such as the Acoustetron II (which we have here at SDSC) costs
upwards of $10K, for just 8 spatialized sources, without room modeling.
I am currently aware of no real-time acoustic modeling system that handles
blocking objects.
Dave Nadeau
San Diego Supercomputer Center