<s>
Perceptual-based	B-Application
3D	I-Application
sound	I-Application
localization	I-Application
is	O
the	O
application	O
of	O
knowledge	O
of	O
the	O
human	O
auditory	O
system	O
to	O
develop	O
3D	O
sound	O
localization	O
technology	O
.	O
</s>
<s>
Emulating	O
the	O
mechanisms	O
of	O
binaural	O
hearing	O
can	O
improve	O
recognition	O
accuracy	O
and	O
signal	O
separation	O
in	O
DSP	B-General_Concept
algorithms	O
,	O
especially	O
in	O
noisy	O
environments	O
.	O
</s>
<s>
While	O
the	O
relationship	O
between	O
human	O
perception	O
of	O
sound	O
and	O
various	O
attributes	O
of	O
the	O
sound	O
field	O
is	O
not	O
yet	O
well	O
understood	O
,	O
DSP	B-General_Concept
algorithms	O
for	O
sound	O
localization	O
are	O
able	O
to	O
employ	O
several	O
mechanisms	O
found	O
in	O
neural	O
systems	O
,	O
including	O
the	O
interaural	O
time	O
difference	O
(	O
ITD	O
,	O
the	O
difference	O
in	O
arrival	O
time	O
of	O
a	O
sound	O
between	O
two	O
locations	O
)	O
,	O
the	O
interaural	O
intensity	O
difference	O
(	O
IID	O
,	O
the	O
difference	O
in	O
intensity	O
of	O
a	O
sound	O
between	O
two	O
locations	O
)	O
,	O
artificial	O
pinnae	O
,	O
the	O
precedence	O
effect	O
,	O
and	O
head-related	O
transfer	O
functions	O
(	O
HRTF	O
)	O
.	O
</s>
<s>
To	O
determine	O
the	O
position	O
of	O
an	O
audio	O
source	O
in	O
3D	O
space	O
,	O
the	O
ear	O
input	O
signals	O
are	O
convolved	B-Language
with	O
the	O
inverses	O
of	O
all	O
possible	O
HRTF	O
pairs	O
,	O
where	O
the	O
correct	O
inverse	O
maximizes	O
cross-correlation	O
between	O
the	O
convolved	B-Language
right	O
and	O
left	O
signals	O
.	O
</s>
