Zoom out Search Issue

IEEE SIGNAL PROCESSING MAGAZINE [103] MARCH 2015
Furthermore, adding the rever-
beration of sources (or the loud-
speaker signals in virtualization of
multichannel loudspeaker signals)
can also improve the realism of the
reproduced sound scene [10]. There-
fore, in virtualization, it is quite com-
mon to use BRIRs [1], [5] that
encapsulate HRTFs and reverbera-
tion. Accordingly, selecting the cor-
rect amount of early reflections as
well as late reverberation is critical to
recreate a faithful sound environ-
ment [1]. In general, the BRIR that
matches the sound environment of the scene or BRIR of a mix-
ing studio are considered to be more suitable [4]. As discussed
in the section “Challenges,” natural sound rendering requires
the accurate reproduction of both the sound sources and the
sound environment. Compared to the virtualization of multi-
channel loudspeaker signals [Figure 2(a)], the latter technique of
virtualizing the source and environment signals [Figure 2(b)] is
more desirable as it is closer to natural listening [6], [8], [9]. These
virtualization techniques can also be incorporated into spatial
audio coding systems, such as binaural cue coding [11], spatial
audio scene coding [5], and directional audio coding [3].
In virtualization, the directions of the sources [or the loud-
speakers in virtualization of multichannel loudspeaker signals as
in Figure 2(a)] have to be calibrated
according to the head movements
(as in natural listening). To fulfill this
need, the HRTFs/BRIRs in the virtu-
alization are updated on the fly based
on these head movements, which are
often tracked by a sensor (e.g., accel-
erometer, gyroscope, camera, etc.).
The latency between the head track-
ing and sound rendering should be
such that the localization accuracy is
not affected [12]. When incorporated
in the virtualization process, such a
head-tracking system can provide
useful dynamic cues to resolve the localization conflicts [1] and
enhance natural sound rendering [10], [12]. It shall be noted that
head tracking is more critical for the directional sources but less
important for the diffuse signals like environment signals and late
reverberation [12]. This is because the perception of diffuse signals
is less affected by head movements.
Recreating the perception of distance of the sources close to
natural listening is another critical aspect in virtualization for nat-
ural sound rendering. However, the challenges in simulating accu-
rate distance perception are numerous. Human beings’ ability to
accurately estimate these distances has long been known to be
poorer compared to our ability to estimate directions, even in the
physical listening space [1]. Virtual listening through headphones
[FIG2] Virtualization of (a) multichannel loudspeaker signals xn
m
^
h
[5], and (b) multiple sources sn
k
^
h
and environment signals
,.anan
LR
^^
hh
,ynyn
LR
^^
hh
is the signal sent to the left and right ear, respectively. Note that head tracking can be used to update the
selected directions of HRTFs/binaural room impulse responses (BRIRs).
M Multichannel Signals
Virtual
Loudspeaker
Positions
Virtual
Source
Positions
Channel 1 x
1
(n)
Channel 2 x
2
(n)
Channel Mx
M
(n)
…
h
x1L
(n)
...
...
y
L
(n)
y
R
(n)
y
L
(n)
a
L
(n)
a
R
(n)
y
R
(n)
Head
Tracking
Head
Tracking
HRTF/BRIR
K Sources
Source 1 s
1
(n)
Source 2 s
2
(n)
Source Ks
K
(n)
…..
...
...
(a)
(b)
h
x2L
(n)
h
xML
(n)
h
s2L
(n)
h
sKL
(n)
h
x1R
(n)
h
x2R
(n)
h
xMR
(n)
h
s1L
(n)
h
s2R
(n)
h
sKR
(n)
h
s1R
(n)
HRTF/BRIR
A PERSONALIZED
LISTENING EXPERIENCE
AND INCORPORATING THE
INFORMATION OF LISTENING
ENVIRONMENT HAVE ALSO
BEEN TRENDS IN THE
HEADPHONE INDUSTRY. THESE
TRENDS SHARE ONE
COMMON OBJECTIVE—
TO RENDER NATURAL
SOUND IN HEADPHONES.
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
q
q
M
M
q
q
M
M
q
M
THE WORLD’S NEWSSTAND
®
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
q
q
M
M
q
q
M
M
q
M
THE WORLD’S NEWSSTAND
®