Information Technology Reference
In-Depth Information
limit the flexibility required by sound design-
ers. Such restrictions have led scientists to find
imaginative alternatives of rendering spatial audio
in synthetic environments-using strategies such
as compact file sizes, low bit rates, client-based
synthetic sound rendering and so forth, without
impacting on perceived sound quality. In this
section, we will review examples of some of the
popular approaches to rendering spatial sound in
games and VR.
In general, left-right source positioning is rela-
tively easy to achieve on both headphone-based
and speaker-based implementations. However,
front-back sound positioning is often less success-
ful, especially when reproduced on headphones.
The primary reason for this is due to the inherent
requirement of the listener to perform head move-
ment when determining the location of sound in
front or back scenarios. A surround-speaker setup
in this respect does not have the same level of dif-
ficulty due to the arrangement of discrete channels
placed in the physical world. In more sophisticated
headphone scenarios, some VR systems incor-
porate head-tracking to allow for the integration
of head orientation and movement in the virtual
scene. Typically, for headphone output, however,
most implementations for computer games and/
or mobile devices use generic filtering processes.
These are especially vulnerable to difficulties
when attempting to externalize the sound in front-
back scenarios. The use of the term “externalize”
in this context relates to the impression the listener
has of the sound being some distance out from
(either in front of, behind, above, or below) their
own listening position. Front-back spatial sound
through headphones usually results in sound
sources being heard
inside
the head rather than
being virtually projected out-the impression of
depth would be realized were the sound sources
virtually projected out.
The result arising from this situation is front-
back confusion, something that can negatively
impact on a listener's experience during gameplay
or VR navigation, where characters or objects are
heard momentarily in a different location from
their visual origin.
In addition to front-back confusion, vertical
sound localization remains a significant challenge
in VR and computer games. Until a commercially
viable method for obtaining individualized HRTF
measurements and accurate real-time processing
of head-tracking is achieved, these issues will
continue to task developers who will have to
rely on simpler approaches. The current method
for obtaining an individual's HRTF is by mea-
suring the right and left Head-Related Impulse
Response (HRIR), which can then be convolved
with any mono signal source. Essentially, the
HRTF is the Fourier Transform of the HRIR.
The HRTF measurements are usually undertaken
in an anechoic chamber with specialized equip-
ment. Most HRTF implementations in computer
games and VR systems are derived from generic
HRTF databases developed from a specialized
human head manikin or derived from an average
set of HRTFs taken from a particular population.
The pinnae and head dimensions of the manikin
head devices are procured from statistical data of
average human biometrics. Of course, the disad-
vantage to this approach is that the player's ear
and head shape may be very different from that
of the manikin, which results not only in the lack
of a true, individualized spatial experience, but in
the distortion of spatial listening cues. However,
recent research into novel ways of acquiring in-
dividualized characteristics of the human hearing
system are being explored, paving the way forward
to exciting developments in spatial audio for VR
and computer games (Satoshi & Suzuki, 2008)
(Otani & Ise, 2003).
Ambisonics
Ambisonics is a spatial audio system that was
developed in the 1970s by Michael Gerzon and
often touted as being superior in terms of spatial
reproduction when compared to commercial do-
mestic multichannel formats. It is a system that
Search WWH ::
Custom Search