to the computer or to the headphones.
When the HRTFs used in processing input
audio are varied in real time in response to
your head turning, the virtual loudspeaker
stays at a constant position within the room,
rather than at a constant angle relative to
your nose. Thus, for DVD playback, when
the center channel loudspeaker is supposed
to be delivering the voice of a character that
you see on the screen, the voice will stay put
at the screen even though you may turn your
head far to the left or right.
Likewise, the position of sound effects
intended to arrive from the rear will be main-
tained for you by making the appropriate
adjustments to the HRTFs used to process the
rear audio channels. What allows the head-
tracking headphone system to make such
clear front/rear distinctions is that you can
turn your ear towards a sound source to fi rmly
establish where it is. If the sound is supposed
to be directly in front of you, then turning your
head to the right will make the sound arrive
earlier at your left ear; but if the sound were to
arrive from the rear, the opposite occurs—the
sound will arrive earlier at your right ear as you
turn your head to the right.
In fact, it’s not absolutely necessary
that the head tracking be done at your own
head (egocentrically, in fi rst-person per-
spective) for you to benefi t from the real-
time adjustment of HRTFs. The adjustment
can be done exocentrically, via your con-
trol of a character’s movement in a gam-
ing scenario, in third-person perspective.
For example, after many twists and turns,
movement through a computer-generated
visual scene might leave you spatially
disoriented. But if through the many turns
you always hear the watery sound of a dis-
tant fountain in a constant location within
the game, then the direction in which you
would need to go in order to return to the
fountain is always clear.
This works for headphone surround in
a wonderful way: The front/rear distinction
here is supported by interactive adjustment
of the HRTF-based audio processing to
maintain a spatially stable mental image of
the game’s virtual world from the perspec-
tive of the character under your control.VIRTUAL SURROUND-SOUND
FROM TWO SPEAKERS
HRTFs can also be used to mix fi ve audio
channels down for playback over two loud-
speakers while maintaining spatial imagery.
These solutions are typically termed “virtual
surround,” and there are various ways to get
a satisfying result.One way is to change the audio repro-
duction scheme so that the loudspeaker
signals reaching your ears are more like
what you get when listening to headphones.
The most straightforward way to do this is to
equalize the sound reproduced at your right
ear so that the sound signal from the right
loudspeaker shows up with a perfectly fl at
frequency response. In effect, you get rid of
your HRTF for the right loudspeaker.
After you’ve done the same for the left
loudspeaker, however, there is still a problem
with reproduction using loudspeakers that
you don’t encounter with headphones: cross-
talk. Crosstalk describes a scenario in which
some of the sound from the right loudspeaker
reaches your left ear, and vice versa. The
technique for eliminating crosstalk is—you
guessed it—crosstalk cancellation, and it
has been practiced for more than 40 years
to improve the loudspeaker reproduction of
“head-related” (aka binaural) recordings.
Two-loudspeaker “virtual surround”
reproduction can also be improved if
coupled with head-tracking technology, but
such solutions are not common. So if you
move your head while listening to “virtual
surround,” you can quickly lose the advan-
tage gained by simple crosstalk cancellation.
There are alternatives that do not require you
to hold yourself very still while listening, butin creating a larger listening area, something
of the surround-sound effect is usually lost.WHAT’S NEXT?
More conventional stereo down-mixing
techniques involve the simple summing
of signals. HRTF-based technology deliv-
ers a surround-sound experience with a
much more adequate rendering of spatial
imagery, while still utilizing only two audio
channels. So what, if anything, is missing?
There is always a need for better content
to showcase what the technology can do,
but there is also room for improvement in
the deployment of HRTFs. Clear distinc-
tions between front and rear virtual source
locations are still diffi cult to achieve using
loudspeakers, and they’re pretty shaky
when headphone reproduction doesn’t
include head tracking.
Because we all have outer ears of dif-
ferent shapes and sizes, it might be that
individually adjusted or customized sets of
HRTFs will provide better solutions for those
users who care to take the time needed to
select their optimum fi lters. Indeed, this is
a hot topic for research and development
of new HRTF-based technology, and it is
expected that new solutions will provide
improved spatial imagery for more and more
users as these solutions are rolled out.Sony’s unique MDR-DS8000 Digital Surround Headphone System uses HRTFs to produce
5.1-channel surround sound using two speakers. The device also features a built-in
gyroscope that tracks the movement of your head so that virtual audio sources stay in
the same position no matter which way your head is turned.William Martens, Ph.D. is an associate professor at McGill University. He is a
perceptual psychologist specializing in spatial hearing research and the simulation
of the acoustical cues used in human sound localization.
BIOr&dBREAKING DOWN TECH —PRESENT AND FUTURE
68 MA XIMUMPC JANUARY 2006
Continued from previous page