Using multiple sounds to add realism to VRML worlds
In the Real World, sound is not round. It doesn't even come in
ellipsoidal packages. It bounces off walls, gets absorbed by carpets
and walls, changes pitch as it moves and gets filtered in some way by
every object in the room. Reproducing, and sometimes even exaggerating
these effects can add a lot to realism of your VRML world, but to do
this, you need to get creative with the VRML sound primitives.
Define the Effect
The first step in adding realism to your soundscape is to determine
what effects will offer the most bang for the byte. Listen carefully to
what happens in the real world when you walk from room to room. Another
good source of ideas is the movie industry, as movie sound effects are
often exagerated. Try to pinpoint the effect that you want to try to
achieve. Here are some ideas to get you started:
-
Reverberation is often present in large rooms or halls.
-
Low frequencies tend to travel better through and around objects. This
leads to a muffled effect on sounds at a distance, or sounds behind
walls or other obstacles, and bright, clear sounds when the listener is
near the source.
-
Echos in a cave or a mountain range are caused by sound reflecting off
of the walls.
Reproduce the effect
The VRML sound primitives don't provide a means for achieving any of
the effects we've just mentioned, so we'll ned to resort to other means
for reproducing or approximating the effect we're after. The two
techniques I've used are:
Sound Editting software - There are a large number of tools on
the Mac, PC's and SGI platforms which will allow you to filter and
manipulate a sampled sound in a variety of ways. Many of these provide
plugin interfaces so you can purchase additional filters and effects.
You can think of these tools as the Photoshop of the sound world. Some
of the examples that follow were processed using the SoundTrack tool
that is bundled with SGI's Irix 6.3 and later.
Black-Box Effect processors - There are is a variety of
black-box sound hardware which will perform real-time sound processing
as well. You can find one of these in most recording studios. While
these black boxes are usually a little less flexible than software
tools, they are often readily available and provide (at least) reverb,
equalization, and pitch effects. Some of the examples that follow were
processed using an Aleisis Quadraverb 2 effect processor.
"Wet" and "Dry" sounds
Once you have a tool which can reproduce the desired effect, you'll
want to make two versions of your sound. One, version should be
"dry" or un-effected. This is the original version of your
sound before you started all this nonsense.
The other version should be the "wet" or processed sound. In
most cases you'll probably be mixing the wet and dry sounds together,
so you'll want your processed sound to contain as little of the
original sound as possible. Here are some examples of wet/dry sound
pairs. We'll use these sounds in the examples that follow.
| Sound Source |
Effect |
|
|
| Courtyard Birds Chirping |
Echo/Reverb |
Dry |
Wet |
| Jungle Island Waterfall |
Equalization |
Dry |
Wet |
| Dance Club Music |
Equalization |
Dry |
Wet |
If the sound is intended to loop, then the wet and the dry sounds
must be the exact same length. Even a slight difference between the
lenghts of the sounds can get them out of phase and ruin the effect.
This part can be tricky, as the "wet" sound generally tends
to be longer. The details of the sound editting technique are beyond
the
Spatial mixing
Now, to get to some VRML authoring! The final step is produce an
effective mix between the wet and dry sounds, so when the
viewer(listener) enters the appropriate environment, the sound supports
their action.
To produce the mix, place 2 Sound nodes, and 2 Audio clips in your
world, carefully adjusting the sound locations and ranges (using the
MinFront/Back and MaxFront/Back fields) to acheive the desired effect.
See the
Sound Node in the VRML
2.0 Spec for more details on how to use these fields.
The exact placement and ranges of the Sound nodes depends on the
effect, but there are 2 things in common to all of the examples
presented here:
-
The wet and dry sounds overlap - The range ellipses defined
by maxFront/maxBack have some area of overlap. Where the range ellipses
overlap, the sounds are mixed together. This is what I like to refer to
as Spatial Mixing.
-
The sounds start at exactly the same time - for looping
sounds, this usually means startTime == 1. Remember, it's important for
looping sounds to have the exact same length, so they don't get out of
phase.
Two of the examples rely only on spatial mixing to perform the wet/dry
mix. The final example uses an interpolator to animate the intensity
field of the "wet" sound in response to a user action.
Examples
Much of the sound and geometry has been removed from this world to ease
up on download times and frame rates. In this world, the two sounds are
located at the same point in space, but have different radii (see
figure). The low rumbling of the waterfall can be heard all throughout
the island, while the actually splashing of the water can only be hear
near the waterfall itself. The parametric equalizer on a personal mixer
was used to separate the two sounds.
Sounds Samples:
-
Low Waterfall Rumble
-
Higher frequency
water splashing
-
Mix heard while standing in close to the
waterfall
 |
Fig. 1 - The two spheres represent the range
ellipses for the wet (inner ellipse)
and dry (outer ellipse) sounds
|
Again, this is a greatly reduced sample out of the Out of Box
Experience. In this example, the ambience of the courtyard gets
reverberated and echoed as the viewer enters either of the two caves.
Great care was taken to get loops of exactly the same length for the
wet and dry sounds. This provides a seamless transition of ambient
sound inside and outside the cave. A small water dripping sound was
added to the wet ambience for added effect.
Sound Samples:
Fig. 2 The sound ellipse for the "wet" sound inside
the East cave.
This example simulates the muffled effect that occurs when sound
travels through walls. Approaching the dance club, you can hear the
beat thumping and the low frequencies of the music. When you open the
door (click on it), the mid and high frequencies of the music are
revealed.
The two sounds are located at the same point in space. The range on the
high frequency sound is slightly smaller than the low frequency sound,
but the intensity field is set to ZERO. This means that the high
frequency sounds will be silent as the viewer approaches.
An interpolator which animates from 0.0 to 1.0 is then wired to the
intensity field of this sound node. The interpolator is triggered by
clicking on the door, so as the door opens, the intensity (or volume)
of the high frequencies increases. This causes the mid and high
frequencies to fade into the mix to reveal the full spectrum of the
music.
The Equalization filter provided with the SoundTrack tool was used to
create the two sound clips in this example.
Sound Samples:
