I
nte
rna
t
io
na
l J
o
urna
l o
f
I
nfo
rm
a
t
ics a
nd
Co
m
m
un
ica
t
io
n T
ec
hn
o
lo
g
y
(
I
J
-
I
CT
)
Vo
l.
15
,
No
.
4
,
Dec
em
b
er
20
26
,
p
p
.
1
6
7
3
~1
6
8
1
I
SS
N:
2252
-
8
7
7
6
,
DOI
:
1
0
.
1
1
5
9
1
/iji
ct
.
v15
i
4
.
pp
1
6
7
3
-
1
6
8
1
1673
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ict.
ia
esco
r
e.
co
m
Accent
-
a
wa
re ara
bic speech
proces
sing
using
CNN
–
BiL
STM
f
o
r
dia
lect
cla
ss
ificat
i
o
n
T
a
rik
Abdel
-
K
a
re
em
AbuAin
1
,
Wa
leed
Abdel K
a
rim
Ab
ua
in
2
1
C
o
l
l
e
g
e
o
f
C
o
m
p
u
t
i
n
g
a
n
d
I
n
f
o
r
m
a
t
i
c
s,
P
r
i
n
c
e
M
u
sai
d
b
i
n
A
b
d
u
l
r
a
h
ma
n
U
n
i
v
e
r
si
t
y
,
R
i
y
a
d
h
,
S
a
u
d
i
A
r
a
b
i
a
2
C
o
l
l
e
g
e
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
T
a
i
b
a
h
U
n
i
v
e
r
s
i
t
y
,
Y
a
n
b
u
,
S
a
u
d
i
A
r
a
b
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Sep
2
7
,
2
0
2
5
R
ev
is
ed
Au
g
1
8
,
2
0
2
6
Acc
ep
ted
Au
g
2
8
,
2
0
2
6
T
h
e
A
ra
b
ic
la
n
g
u
a
g
e
p
re
se
n
ts c
o
n
si
d
e
ra
b
le
d
ia
lec
ta
l
d
i
v
e
r
si
t
y
,
w
h
ic
h
c
a
n
p
o
s
e
c
h
a
ll
e
n
g
e
s
f
o
r
s
p
e
e
c
h
tec
h
n
o
l
o
g
ie
s,
p
a
rt
ic
u
la
rl
y
a
u
t
o
m
a
ti
c
s
p
e
e
c
h
r
e
c
o
g
n
it
i
o
n
(AS
R)
.
Tra
d
it
i
o
n
a
l
m
o
d
e
ls
tra
i
n
e
d
p
r
ima
ri
l
y
o
n
m
o
d
e
r
n
s
ta
n
d
a
rd
Ara
b
ic
(M
S
A
)
d
o
n
o
t
a
c
c
o
u
n
t
f
o
r
re
g
i
o
n
a
l
a
c
c
e
n
ts
,
lea
d
in
g
t
o
p
o
o
r
p
e
r
f
o
r
m
a
n
c
e
o
n
u
n
d
e
r
re
p
re
se
n
te
d
d
ia
lec
ts.
H
o
w
e
v
e
r,
t
h
is
p
a
p
e
r
a
d
d
r
e
ss
e
s
t
h
is
g
a
p
b
y
in
t
ro
d
u
c
i
n
g
a
h
y
b
r
i
d
CNN
-
Bi
L
S
TM
m
o
d
e
l
d
e
si
g
n
e
d
f
o
r
a
u
t
o
m
a
t
ic
A
ra
b
ic
d
i
a
lec
t
c
las
s
i
fica
ti
o
n
.
T
h
e
d
a
tas
e
t
wa
s
d
i
v
i
d
e
d
i
n
t
o
fi
v
e
c
a
te
g
o
rie
s
g
a
t
h
e
r
e
d
fr
o
m
v
a
r
i
o
u
s
Y
o
u
T
u
b
e
v
id
e
o
s:
Le
v
a
n
ti
n
e
(
J
o
r
d
a
n
,
Le
b
a
n
o
n
,
S
y
r
ia
,
P
a
l
e
st
i
n
e
)
,
G
u
lf
(
S
a
u
d
i
Ar
a
b
ia
,
Ye
m
e
n
,
Q
a
tar
,
UA
E)
,
N
il
e
Va
l
le
y
(E
g
y
p
t
,
S
u
d
a
n
)
,
M
e
s
o
p
o
t
a
m
ia
n
(
Ira
q
,
K
u
wa
it
),
a
n
d
N
o
r
t
h
A
fr
ica
n
(A
l
g
e
r
ia
,
L
i
b
y
a
,
M
a
u
r
it
a
n
ia
,
M
o
r
o
c
c
o
).
T
h
e
re
f
o
re
,
a
s
y
ste
m
a
t
i
c
a
u
g
m
e
n
tat
i
o
n
wa
s
a
p
p
l
ie
d
t
o
a
d
d
re
ss
t
h
is
im
b
a
la
n
c
e
,
i
n
c
l
u
d
i
n
g
p
it
c
h
sh
if
t
i
n
g
,
ti
m
e
st
re
t
c
h
i
n
g
,
a
n
d
a
m
p
li
t
u
d
e
sc
a
l
in
g
,
lea
d
i
n
g
t
o
4
,
0
0
0
b
a
la
n
c
e
d
u
tt
e
ra
n
c
e
s
st
a
n
d
a
r
d
i
z
e
d
t
o
1
6
k
Hz
a
n
d
8
se
c
o
n
d
s.
Co
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
tw
o
r
k
(
CNN
)
la
y
e
r
s
c
a
p
t
u
r
e
d
l
o
c
a
li
z
e
d
s
p
e
c
tra
l
fe
a
t
u
re
s
,
w
h
il
e
b
i
d
i
re
c
ti
o
n
a
l
l
o
n
g
s
h
o
r
t
-
te
rm
m
e
m
o
r
y
(
B
iL
S
T
M
)
u
n
it
s
m
o
d
e
le
d
l
o
n
g
-
ra
n
g
e
tem
p
o
ra
l
d
e
p
e
n
d
e
n
c
ies
.
E
v
a
l
u
a
t
i
o
n
m
e
t
ric
s,
in
c
l
u
d
i
n
g
c
las
s
if
ica
ti
o
n
re
p
o
r
t,
c
o
n
f
u
s
i
o
n
m
a
tr
i
x
,
a
n
d
l
e
a
r
n
i
n
g
c
u
r
v
e
s
,
s
h
o
we
d
e
ffe
c
t
i
v
e
p
e
r
f
o
rm
a
n
c
e
,
w
it
h
a
n
o
v
e
ra
ll
a
c
c
u
ra
c
y
o
f
8
8
%
.
K
ey
w
o
r
d
s
:
Ar
ab
ic
ac
ce
n
ts
B
iLST
M
C
NN
Dialec
t
class
if
icatio
n
Sp
ee
ch
p
r
o
ce
s
s
in
g
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
T
ar
ik
Ab
d
el
-
Kar
ee
m
Ab
u
Ain
C
o
lleg
e
o
f
C
o
m
p
u
tin
g
an
d
I
n
f
o
r
m
atics
,
Sau
d
i E
lectr
o
n
ic
U
n
iv
er
s
ity
R
iy
ad
h
1
1
6
7
3
,
Sau
d
i A
r
ab
ia
E
m
ail: t.
ab
o
ain
@
s
eu
.
ed
u
.
s
a
1.
I
NT
RO
D
UCT
I
O
N
Ar
ab
ic
,
s
p
o
k
en
b
y
4
2
0
m
illi
o
n
p
eo
p
le
wo
r
l
d
wid
e
[
1
]
,
h
as
a
v
ar
iety
o
f
d
ialec
ts
ac
r
o
s
s
r
eg
io
n
s
,
with
s
o
m
e
d
ialec
ts
o
f
ten
s
u
b
s
tan
tially
d
if
f
er
en
t
f
r
o
m
o
th
er
s
[
2
]
.
Ar
ab
ic
ac
ce
n
ts
class
if
icatio
n
i
s
co
m
p
lex
b
ec
au
s
e
th
e
lan
g
u
ag
e
e
n
co
m
p
ass
es
m
an
y
d
ialec
ts
th
r
o
u
g
h
o
u
t
th
e
A
r
ab
wo
r
ld
.
W
h
ile
m
o
d
er
n
s
tan
d
ar
d
Ar
a
b
ic
(
MSA)
s
er
v
es
as
th
e
f
o
r
m
al
w
r
itten
an
d
s
p
o
k
e
n
s
tan
d
ar
d
u
s
ed
i
n
m
ed
ia
a
n
d
ed
u
ca
tio
n
,
n
ati
v
e
s
p
ea
k
er
s
m
ain
ly
co
m
m
u
n
icate
in
v
er
n
ac
u
lar
s
t
h
at
v
ar
y
s
ig
n
if
ican
tly
in
p
r
o
n
u
n
ciatio
n
,
v
o
ca
b
u
lar
y
,
an
d
s
y
n
t
ax
.
T
h
ese
ar
e
d
ialec
tal
v
ar
iatio
n
s
,
n
o
r
m
ally
r
o
o
te
d
in
g
e
o
g
r
a
p
h
y
,
h
is
to
r
y
,
an
d
cu
ltu
r
e,
r
esu
ltin
g
in
a
r
an
g
e
o
f
ac
ce
n
ts
th
at
ca
n
p
r
ev
en
t
m
u
tu
al
u
n
d
er
s
tan
d
in
g
,
ev
e
n
a
m
o
n
g
n
ativ
e
s
p
ea
k
er
s
.
Fo
r
in
s
tan
ce
,
p
r
o
n
u
n
ciatio
n
d
if
f
er
e
n
ce
s
b
et
wee
n
Gu
lf
d
ialec
ts
an
d
Ma
g
h
r
eb
i
Ar
ab
ic
ca
n
s
ig
n
if
ican
tl
y
af
f
ec
t
clar
ity
a
n
d
m
ea
n
in
g
.
T
h
is
lin
g
u
is
tic
d
iv
e
r
s
ity
p
o
s
es
a
ch
allen
g
e
f
o
r
s
p
ee
ch
-
b
ased
tech
n
o
lo
g
ies,
esp
ec
ially
au
to
m
atic
s
p
ee
ch
r
ec
o
g
n
itio
n
(
ASR
)
,
wh
ich
s
tr
u
g
g
les to
p
e
r
f
o
r
m
well
ac
r
o
s
s
s
u
ch
v
ar
ied
lan
g
u
ag
e
g
r
o
u
p
s
[
3
]
.
Stan
d
ar
d
ASR
m
o
d
els,
p
r
im
ar
ily
d
ev
elo
p
e
d
f
o
r
n
ativ
e
s
p
ee
ch
[
4
]
an
d
tr
ain
ed
o
n
MSA
o
r
lim
ited
-
ac
ce
n
t
v
a
r
ieties,
o
f
ten
s
tr
u
g
g
le
to
ac
cu
r
ately
in
ter
p
r
et
o
r
d
etec
t
s
p
ee
c
h
f
r
o
m
less
-
co
m
m
o
n
d
ialec
ts
.
Acc
en
t
class
if
icatio
n
aim
s
to
tack
le
th
is
is
s
u
e
b
y
au
t
o
m
at
ically
id
en
tify
in
g
th
e
s
p
ea
k
e
r
’
s
ac
ce
n
t,
en
ab
lin
g
ASR
an
d
o
th
er
s
p
ee
ch
s
y
s
tem
s
to
ad
ap
t in
r
ea
l tim
e.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
4
,
Dec
em
b
er
20
26
:
1
6
7
3
-
1
6
8
1
1674
Ap
ar
t
f
r
o
m
ac
h
ie
v
i
n
g
h
ig
h
er
r
ec
o
g
n
itio
n
ac
c
u
r
ac
y
,
th
is
p
r
o
ce
s
s
is
n
ee
d
ed
to
cr
ea
te
in
clu
s
iv
e,
cu
ltu
r
ally
c
o
n
s
cio
u
s
AI
s
y
s
tem
s
th
at
b
etter
s
er
v
e
A
r
ab
ic
-
s
p
ea
k
in
g
p
o
p
u
latio
n
s
.
Sp
ec
if
ic
ally
,
o
u
r
m
o
d
el
u
s
es
lab
el
s
m
o
o
th
in
g
,
a
f
ix
ed
2
4
×
3
2
s
p
ec
tr
o
g
r
am
in
p
u
t,
an
d
ad
ap
tiv
e
lear
n
in
g
r
ate
s
ch
ed
u
lin
g
,
d
esig
n
ch
o
ices
th
at
ad
d
r
ess
th
e
ac
o
u
s
tic
v
ar
iab
ilit
y
o
f
Ar
ab
ic
d
ialec
ts
.
Un
lik
e
s
in
g
le
-
s
tr
ea
m
co
n
v
o
lu
tio
n
a
l
n
eu
r
al
n
etwo
r
k
s
(
C
NNs
)
o
r
B
i
GR
U
m
o
d
els,
C
NN
-
B
iLST
M
jo
in
tly
ca
p
tu
r
es
lo
ca
l
s
p
ec
tr
al
p
atter
n
s
an
d
lo
n
g
-
r
an
g
e
tem
p
o
r
al
d
ep
en
d
e
n
cies,
t
h
er
eb
y
ac
h
ie
v
in
g
s
u
p
er
i
o
r
d
ialec
t
d
if
f
e
r
e
n
tiatio
n
.
I
n
th
is
p
a
p
er
,
we
in
tr
o
d
u
ce
a
h
y
b
r
id
C
NN
-
B
iL
STM
m
eth
o
d
f
o
r
c
lass
if
y
in
g
lig
h
t
Ar
ab
ic
ac
ce
n
ts
.
T
h
e
d
ata
b
ase
is
s
p
lit
in
to
f
iv
e
ca
teg
o
r
ies:
L
ev
an
tin
e
(
J
o
r
d
a
n
,
L
eb
a
n
o
n
,
Sy
r
ia,
Palest
in
e)
,
Gu
lf
(
Sau
d
i
Ar
ab
ia,
Yem
en
,
Qata
r
,
UAE
)
,
Nile
Valley
(
Su
d
an
,
E
g
y
p
t)
,
Me
s
o
p
o
tam
ia
n
(
I
r
aq
an
d
Ku
wait)
,
an
d
No
r
t
h
Af
r
ican
(
Alg
er
ia,
L
ib
y
a,
Ma
u
r
itan
ia,
Mo
r
o
cc
o
)
.
T
h
e
r
esu
lts
o
b
tain
ed
will b
e
a
n
aly
ze
d
in
t
h
e
p
r
o
v
id
ed
s
ec
tio
n
s
.
2.
RE
L
AT
E
D
WO
RK
I
n
r
ec
en
t
y
ea
r
s
,
m
an
y
r
esear
ch
er
s
h
av
e
c
o
n
d
u
cted
s
tu
d
ies
o
n
Ar
ab
ic
d
ialec
t
id
e
n
tific
atio
n
.
I
n
th
is
s
ec
tio
n
,
we
will
r
ev
ie
w
s
o
m
e
o
f
th
ese
s
tu
d
ies
to
g
ain
in
s
ig
h
t
in
to
th
e
m
eth
o
d
s
an
d
m
o
d
els
p
r
ev
i
o
u
s
ly
u
s
ed
in
th
is
ar
ea
.
Aliwy
et
a
l.
[
5
]
co
n
d
u
cted
a
s
tu
d
y
th
at
ap
p
lied
p
r
ep
r
o
ce
s
s
in
g
,
clu
s
ter
in
g
,
an
d
en
s
em
b
le
class
if
ier
s
to
th
e
NADI
d
ataset,
r
ea
ch
in
g
an
F
-
m
ea
s
u
r
e
o
f
as
h
ig
h
as
5
2
.
3
8
%.
No
tab
ly
,
clu
s
ter
in
g
p
lay
e
d
a
k
ey
r
o
le
i
n
s
ig
n
if
ican
tly
e
n
h
an
cin
g
th
e
ac
cu
r
ac
y
o
f
Ar
ab
ic
d
ialec
t c
lass
if
icatio
n
.
E
lm
ah
d
y
et
a
l.
[
6
]
p
r
esen
t
a
n
o
v
el
c
r
o
s
s
-
lin
g
u
al
a
d
ap
t
atio
n
ap
p
r
o
ac
h
(
ML
L
R
/MAP)
th
at
lev
er
ag
es
MSA
s
p
ee
ch
r
eso
u
r
ce
s
to
en
h
an
ce
ASR
f
o
r
d
ialec
tal
Ar
ab
ic.
T
h
e
MSA
d
ataset
co
m
p
r
is
es
3
3
h
o
u
r
s
,
E
C
A
h
as
0
.
5
h
o
u
r
s
,
a
n
d
L
C
A
in
clu
d
es
2
6
h
o
u
r
s
.
R
esu
lts
in
d
icate
a
4
1
.
8
%
r
elativ
e
r
ed
u
ctio
n
in
W
E
R
o
n
E
C
A
with
m
o
d
if
ied
p
h
o
n
em
ic
m
o
d
els.
T
h
e
m
o
d
el
ac
h
iev
ed
9
1
.
3
4
%
ac
cu
r
ac
y
,
s
u
r
p
ass
in
g
b
aselin
es
s
u
ch
as
R
NN
(
7
6
.
4
1
%)
an
d
C
NN
(
8
1
.
4
3
%).
T
ab
le
1
p
r
o
v
id
es
an
o
v
e
r
v
ie
w
o
f
r
ec
e
n
t
s
tu
d
ies
o
n
th
e
clas
s
if
icatio
n
o
f
wr
itten
Ar
ab
ic
d
ialec
ts
.
Als
ad
h
an
[
7
]
ap
p
lied
Ar
aBER
T
an
d
Fas
tTe
x
t/DB
S
C
AN
to
s
o
cial
m
ed
ia
d
ata,
attain
in
g
8
8
.
9
% d
ialec
t c
lass
if
icatio
n
ac
cu
r
ac
y
an
d
ef
f
ec
tiv
e
em
o
tio
n
d
etec
tio
n
.
T
ab
le
1
.
R
ec
en
t stu
d
ies o
f
wr
it
ten
Ar
ab
ic
d
ialec
t c
lass
if
icatio
n
R
e
f
.
M
e
t
h
o
d
D
a
t
a
s
e
t
s
F
i
n
d
i
n
g
s
Y
a
f
o
o
z
[
8
]
R
N
N
s (LS
T
M
/
G
R
U
)
,
Tr
a
n
sf
o
r
m
e
r
s
(
A
r
a
B
ER
T
,
a
n
d
M
A
R
B
E
R
T
v2
)
,
a
n
d
L
LM
s
(
G
e
mi
n
i
,
Q
w
e
n
)
Q
A
D
I
(
4
4
0
K
A
r
a
b
i
c
t
w
e
e
t
s
,
1
8
d
i
a
l
e
c
t
s)
Th
e
a
c
h
i
e
v
e
d
a
c
c
u
r
a
c
y
i
s
-
6
5
%
w
i
t
h
M
A
R
B
ER
Tv
2
,
a
n
d
5
2
%
w
i
t
h
R
N
N
s
El
saft
y
e
t
a
l
.
[
9
]
A
r
D
i
a
B
E
R
T/
A
r
D
i
a
G
P
T
w
i
t
h
r
a
r
e
-
d
i
a
l
e
c
t
b
a
l
a
n
c
i
n
g
A
r
D
i
a
(
9
0
3
k
t
e
x
t
s,
6
d
i
a
l
e
c
t
s,
l
o
n
g
-
f
o
r
m T
i
k
To
k
/
W
i
k
i
p
e
d
i
a
d
o
m
i
n
a
n
c
e
)
Th
e
me
t
h
o
d
o
b
t
a
i
n
e
d
8
6
.
5
5
%
M
a
c
r
o
-
F1
S
a
l
e
h
e
t
a
l
.
[
1
0
]
h
y
b
r
i
d
s
t
a
c
k
e
d
t
r
a
n
sf
o
r
mer m
o
d
e
l
s
S
h
a
m
i
a
n
d
I
A
D
D
Th
e
a
c
h
i
e
v
e
d
a
c
c
u
r
a
c
y
i
s
-
S
h
a
m
i
:
8
9
.
7
3
%
a
n
d
A
D
D
:
9
3
.
0
6
2
%
A
b
d
u
l
l
a
h
e
t
a
l
.
[
11
]
V
o
i
c
e
C
o
n
v
e
r
si
o
n
(
V
C
)
-
E
n
h
a
n
c
e
d
Tr
a
i
n
i
n
g
M
G
B
-
3
A
D
I
-
5
(
5
3
.
6
h
r
s,
1
4
.
6
k
samp
l
e
s)
f
o
r
t
r
a
i
n
i
n
g
a
n
d
M
A
D
I
S
-
5
(
c
u
r
a
t
e
d
1
2
-
h
r
c
r
o
ss
-
d
o
m
a
i
n
s
e
t
)
f
o
r
e
v
a
l
u
a
t
i
o
n
Th
e
a
p
p
r
o
a
c
h
y
i
e
l
d
e
d
c
o
n
si
s
t
e
n
t
i
m
p
r
o
v
e
me
n
t
s
o
f
+
3
4
.
1
%
A
l
h
a
r
b
i
[
12
]
A
n
o
v
e
l
M
u
l
t
i
m
o
d
a
l
F
r
a
m
e
w
o
r
k
,
c
a
l
l
e
d
M
F
-
S
a
u
d
i
S
a
u
d
i
D
i
a
l
e
c
t
A
u
d
i
o
-
Te
x
t
Th
e
a
c
c
u
r
a
c
y
a
c
h
i
e
v
e
d
i
s
9
6
.
5
%
A
l
su
w
a
y
l
i
m
i
[
13
]
H
y
b
r
i
d
C
A
M
e
L
B
E
R
T/
A
L
B
E
R
T
+
B
i
LST
M
;
M
L/
D
L/
t
r
a
n
sf
o
r
mer
3
8
,
3
9
4
a
n
n
o
t
a
t
e
d
s
o
c
i
a
l
m
e
d
i
a
c
o
mm
e
n
t
s (
4
d
i
a
l
e
c
t
s)
Th
e
a
c
h
i
e
v
e
d
a
c
c
u
r
a
c
y
i
s
-
C
A
M
e
LB
ER
T
:
8
7
.
6
7
%
a
n
d
A
LB
ER
T:
8
6
.
5
1
%
T
ib
i
an
d
Me
s
s
ao
u
d
[
1
4
]
cr
e
ated
a
s
y
s
tem
th
at
d
etec
ts
Ar
ab
ic
d
ialec
ts
b
y
ap
p
ly
in
g
m
u
lti
-
s
ca
le
p
r
o
d
u
ct
an
aly
s
is
(
MPA)
to
e
x
tr
ac
t
f
ea
tu
r
es
f
o
r
a
h
am
ilto
n
n
eu
r
al
n
etwo
r
k
(
HNN)
.
I
t
was
tr
ain
ed
o
n
eig
h
t
L
DC
an
d
eig
h
t
ADI
d
ialec
ts
,
ac
h
iev
in
g
a
n
ac
cu
r
ac
y
o
f
9
1
.
3
4
%,
wh
ich
is
b
etter
th
an
t
h
e
R
NN
(
7
6
.
4
1
%)
a
n
d
C
NN
(
8
1
.
4
3
%)
m
o
d
els.
R
esear
ch
p
r
o
p
o
s
ed
b
y
Ali
[
1
5
]
d
ev
el
o
p
ed
an
en
d
-
to
-
en
d
C
N
N
-
R
NN
-
b
ased
m
o
d
el
with
a
1
,
4
0
0
-
h
o
u
r
m
u
lti
-
d
ialec
t
co
r
p
u
s
(
MG
B
-
2
,
T
ash
k
ee
la,
an
d
s
o
o
n
)
.
I
t
h
ad
a
wo
r
d
er
r
o
r
r
ate
(
W
E
R
)
o
f
1
4
%,
wh
ich
is
h
ig
h
er
th
an
th
at
o
f
p
r
e
v
io
u
s
s
y
s
tem
s
.
A
s
tu
d
y
b
y
Al
-
Ay
y
o
u
b
et
a
l
.
[
1
6
]
u
s
ed
wav
elet
f
ea
tu
r
es
an
d
a
B
ay
esian
Netwo
r
k
cl
ass
if
ier
to
d
if
f
er
en
tiate
b
etwe
en
J
o
r
d
a
n
ian
an
d
E
g
y
p
tian
Ar
ab
ic
d
ialec
ts
.
I
t
was
tr
ai
n
ed
o
n
a
s
m
all
d
ataset
(
a
cu
s
to
m
d
ataset
o
f
3
5
s
am
p
l
es)
an
d
,
u
s
in
g
t
h
ese
f
ea
tu
r
es,
ac
h
iev
ed
9
7
%
ac
cu
r
ac
y
.
T
ab
l
e
2
p
r
e
s
en
ts
r
ec
en
t
s
tu
d
ies
o
n
th
e
class
if
icatio
n
o
f
Ar
ab
ic
s
p
o
k
en
d
ialec
ts
.
Al
-
An
zi
an
d
T
h
a
n
k
alee
la
[
1
7
]
u
s
ed
C
NNs
an
d
MF
C
C
o
n
4
,
0
7
1
A
r
ab
ic
au
d
io
clip
s
ac
r
o
s
s
5
d
ialec
ts
,
ac
h
iev
in
g
9
4
.
2
8
% test ac
cu
r
ac
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
cc
en
t
-
a
w
a
r
e
a
r
a
b
ic
s
p
ee
ch
p
r
o
ce
s
s
in
g
u
s
in
g
C
N
N
–
B
i
LS
T
M fo
r
d
ia
lect
…
(
Ta
r
ik
A
b
d
el
-
K
a
r
ee
m
A
b
u
A
in
)
1675
T
ab
le
2
.
R
ec
en
t stu
d
ies o
f
Ar
a
b
ic
s
p
o
k
en
d
ialec
ts
class
if
icati
o
n
Ref
M
e
t
h
o
d
D
a
t
a
s
e
t
s
F
i
n
d
i
n
g
s
A
l
sah
e
e
l
e
t
a
l
.
[
1
8
]
B
i
G
R
U
a
n
d
B
i
LST
M
S
A
D
A
c
o
r
p
u
s
o
f
d
i
a
l
e
c
t
a
l
A
r
a
b
i
c
-
9
7
%
t
r
a
i
n
i
n
g
a
c
c
u
r
a
c
y
a
n
d
9
0
%
t
e
st
a
c
c
u
r
a
c
y
Al
-
S
h
w
a
y
y
a
t
a
n
d
G
e
r
e
k
[
19
]
H
y
b
r
i
d
M
F
C
C
+
C
N
N
a
n
d
W
a
v
e
l
e
t
+
R
N
N
C
o
mm
o
n
V
o
i
c
e
A
r
a
b
i
c
1
2
.
0
A
c
h
i
e
v
e
d
a
c
c
u
r
a
c
y
:
-
9
1
.
2
%
f
o
r
M
F
C
C
+
C
N
N
v
s.
-
6
6
.
5
%
f
o
r
W
a
v
e
l
e
t
+
R
N
N
Ta
l
a
f
h
a
e
t
a
l
.
[
20
]
Pre
-
t
r
a
i
n
e
d
m
o
d
e
l
s (W
h
i
s
p
e
r
,
w
2
v
-
B
ER
T,
C
o
n
f
o
r
mer)
S
p
e
e
c
h
c
o
r
p
u
s (
C
a
s
a
b
l
a
n
c
a
)
c
o
v
e
r
i
n
g
8
A
r
a
b
i
c
d
i
a
l
e
c
t
s
f
o
r
I
D
/
A
S
R
;
mu
l
t
i
-
s
o
u
r
c
e
c
o
l
l
e
c
t
i
o
n
f
o
r
d
i
a
c
r
i
t
i
z
a
t
i
o
n
Th
e
b
e
st
a
c
h
i
e
v
e
d
a
c
c
u
r
a
c
y
i
s
7
9
.
8
%
B
a
r
a
k
a
t
e
t
a
l
.
[
21
]
A
d
e
e
p
l
e
a
r
n
i
n
g
c
l
a
ssi
f
i
c
a
t
i
o
n
mo
d
e
l
M
o
r
e
t
h
a
n
6
6
,
0
0
0
a
u
d
i
o
f
i
l
e
s
f
r
o
m t
h
e
Y
o
u
T
u
b
e
A
D
I
-
17
M
I
T
d
a
t
a
s
e
t
.
Th
e
a
c
c
u
r
a
c
y
a
c
h
i
e
v
e
d
i
s
8
6
%
o
n
t
e
st
d
a
t
a
,
a
n
d
9
5
%
w
h
e
n
c
o
n
s
i
d
e
r
i
n
g
t
h
e
t
o
p
3
p
r
e
d
i
c
t
e
d
d
i
a
l
e
c
t
s
N
a
h
a
r
e
t
a
l
.
[
2
2
]
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
m
o
d
e
l
s
(
K
-
N
e
a
r
e
st
N
e
i
g
h
b
o
r
(
K
N
N
)
,
R
a
n
d
o
m F
o
r
e
s
t
(
R
F
)
,
M
u
l
t
i
-
La
y
e
r
P
e
r
c
e
p
t
r
o
n
(
M
LP)
,
a
n
d
A
r
t
i
f
i
c
i
a
l
N
e
u
r
a
l
N
e
t
w
o
r
k
s
(ANN))
Th
e
d
a
t
a
se
t
w
a
s
c
o
l
l
e
c
t
e
d
f
r
o
m Y
o
u
T
u
b
e
v
i
d
e
o
s
a
n
d
a
u
d
i
o
s
e
g
m
e
n
t
s
l
a
b
e
l
e
d
b
y
h
u
m
a
n
e
x
p
e
r
t
s
-
M
F
C
C
F
e
a
t
u
r
e
s
K
N
N
:
6
2
%,
R
F
:
6
0
%
M
LP:
3
3
%,
A
N
N
:
4
2
%
-
TFCC
F
e
a
t
u
r
e
s
K
N
N
:
7
6
%,
R
F
:
6
4
%
M
LP:
3
4
%,
A
N
N
:
4
1
%
H
a
mi
d
i
e
t
a
l
.
[
2
3
]
H
M
M
-
G
M
M
w
i
t
h
M
F
C
C
2
4
sp
e
a
k
e
r
s (
M
o
r
o
c
c
a
n
/
A
l
g
e
r
i
a
n
)
-
B
e
st
a
c
c
u
r
a
c
y
:
7
8
.
8
%
w
i
t
h
mi
x
e
d
d
i
c
t
i
o
n
a
r
y
Al
-
A
z
a
n
i
a
n
d
El
-
A
l
f
y
[
24
]
C
o
m
b
i
n
a
t
i
o
n
o
f
a
u
d
i
o
a
n
d
t
e
x
t
u
a
l
f
e
a
t
u
r
e
s f
o
r
A
r
a
b
i
c
d
i
a
l
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
5
2
4
o
p
i
n
i
o
n
u
t
t
e
r
a
n
c
e
s
e
x
t
r
a
c
t
e
d
f
r
o
m
6
3
Y
o
u
T
u
b
e
v
i
d
e
o
s
-
S
p
e
a
k
e
r
-
i
n
d
e
p
e
n
d
e
n
t
:
t
e
x
t
o
u
t
p
e
r
f
o
r
ms
a
u
d
i
o
w
i
t
h
F
1
s
c
o
r
e
o
f
6
3
.
6
1
%
-
S
p
e
a
k
e
r
-
d
e
p
e
n
d
e
n
t
:
b
o
t
h
t
e
x
t
a
n
d
a
u
d
i
o
p
e
r
f
o
r
me
d
si
mi
l
a
r
l
y
,
w
i
t
h
a
n
F
1
sc
o
r
e
o
f
8
5
.
5
2
%
M
a
l
mas
i
a
n
d
Za
mp
i
e
r
[
2
5
]
En
se
mb
l
e
c
l
a
ssi
f
i
e
r
s
Eg
y
p
t
i
a
n
,
G
u
l
f
,
Le
v
a
n
t
i
n
e
,
N
o
r
t
h
A
f
r
i
c
a
n
,
a
n
d
M
S
A
-
Th
e
a
c
h
i
e
v
e
d
a
c
c
u
r
a
c
y
i
s
a
b
o
u
t
6
8
%
3.
M
E
T
H
O
DS A
ND
M
A
T
E
R
I
AL
S
3
.
1
.
Da
t
a
s
et
s
T
h
is
s
tu
d
y
co
llected
a
d
ataset
f
r
o
m
v
ar
i
o
u
s
Yo
u
T
u
b
e
v
id
eo
s
,
ca
teg
o
r
izin
g
it
in
to
f
iv
e
Ar
a
b
ic
d
ialec
t
g
r
o
u
p
s
:
L
ev
an
tin
e
(
J
o
r
d
an
,
L
e
b
an
o
n
,
Sy
r
ia,
Palest
in
e)
,
Gu
lf
(
Sau
d
i
Ar
ab
ia,
Yem
en
,
Qata
r
,
UAE
)
,
Nile
Valley
(
E
g
y
p
t,
Su
d
an
)
,
Me
s
o
p
o
tam
ia
n
(
I
r
a
q
,
Ku
wait)
,
a
n
d
No
r
th
A
f
r
ican
(
Alg
er
ia,
L
ib
y
a,
Ma
u
r
it
an
ia,
Mo
r
o
cc
o
)
,
as
s
h
o
wn
in
Fig
u
r
e
1
.
Yo
u
T
u
b
e
was
s
elec
ted
b
ec
au
s
e
it
o
f
f
e
r
s
n
atu
r
al,
s
p
o
n
tan
eo
u
s
s
p
ee
ch
s
am
p
les,
an
d
th
ese
g
r
o
u
p
s
alig
n
with
wid
ely
ac
ce
p
ted
r
eg
io
n
al
d
ia
lect
class
if
icatio
n
s
.
E
ac
h
ca
teg
o
r
y
co
n
tain
s
lab
eled
au
d
io
s
am
p
les
in
.
wav
f
o
r
m
at
an
d
allo
ws
s
u
p
er
v
is
ed
d
ialec
t
class
if
icatio
n
.
T
h
e
s
tr
u
ctu
r
e
o
f
th
e
d
ataset
f
ac
ilit
ates
r
eg
io
n
al
g
en
er
aliza
tio
n
an
d
aid
s
in
e
f
f
ec
tiv
e
m
o
d
el
tr
ai
n
in
g
an
d
ev
alu
atio
n
.
Ad
d
itio
n
ally
,
th
e
d
ataset
is
d
iv
id
ed
in
to
tr
ain
in
g
an
d
t
esti
n
g
s
ets
(
8
0
/2
0
)
f
o
r
m
o
d
el
d
e
v
elo
p
m
e
n
t.
3
.
2
.
P
re
pro
ce
s
s
ing
da
t
a
a
ug
m
ent
a
t
io
n
I
n
th
is
p
ap
er
,
a
s
y
s
tem
atic
p
r
ep
r
o
ce
s
s
in
g
an
d
au
g
m
en
tati
o
n
m
eth
o
d
f
o
r
Ar
ab
ic
s
p
ee
c
h
d
ata
was
d
ev
elo
p
e
d
to
b
alan
ce
an
d
en
r
ich
th
e
Ar
a
b
ic
s
p
ee
ch
co
r
p
u
s
ac
r
o
s
s
f
iv
e
d
ialec
t
g
r
o
u
p
s
:
L
ev
an
tin
e,
Gu
l
f
,
Nile
Valley
,
Me
s
o
p
o
tam
ian
,
an
d
No
r
th
Af
r
ican
.
All
s
o
u
n
d
f
il
es
wer
e
d
o
wn
s
am
p
le
d
to
1
6
k
Hz
an
d
tim
e
-
n
o
r
m
alize
d
t
o
8
s
ec
o
n
d
s
(
1
2
8
,
0
0
0
s
am
p
les)
to
p
r
o
v
id
e
t
em
p
o
r
al
co
n
s
is
ten
cy
d
u
r
i
n
g
m
o
d
el
tr
ain
in
g
.
T
h
e
t
r
i
m
m
i
n
g
s
w
er
e
p
e
r
f
o
r
m
ed
at
2
0
d
B
,
a
n
d
p
a
d
d
i
n
g
o
r
d
e
let
io
n
s
w
er
e
e
m
p
l
o
y
e
d
t
o
m
ai
n
t
ai
n
u
n
if
o
r
m
l
en
g
t
h
.
I
n
o
r
d
er
t
o
a
d
d
r
ess
d
atas
et
im
b
a
la
n
c
e
an
d
g
e
n
e
r
a
liz
ati
o
n
,
e
ac
h
d
i
ale
ct
al
g
r
o
u
p
w
as
b
a
la
n
c
ed
t
o
8
0
0
u
tt
er
an
ce
s
.
W
h
e
n
t
h
e
a
v
a
il
ab
l
e
r
ec
o
r
d
i
n
g
s
we
r
e
i
n
s
u
f
f
i
cie
n
t
r
ela
ti
v
e
t
o
th
e
ta
r
g
et
,
au
g
m
e
n
ta
ti
o
n
w
as
u
s
e
d
.
An
au
g
m
en
tatio
n
s
tep
in
v
o
lv
e
d
p
itch
s
h
if
tin
g
(
in
cr
ea
s
in
g
o
r
d
ec
r
ea
s
in
g
th
e
p
itch
b
y
1
–
2
s
em
ito
n
es),
s
p
ee
d
v
ar
iatio
n
(
s
lo
win
g
d
o
wn
o
r
s
p
ee
d
in
g
u
p
th
e
a
u
d
io
b
y
u
p
to
1
0
%),
an
d
am
p
litu
d
e
s
ca
lin
g
(
r
ed
u
cin
g
o
r
a
m
p
lify
in
g
th
e
v
o
lu
m
e
b
y
u
p
to
2
0
%),
wh
ic
h
was
ap
p
lie
d
r
a
n
d
o
m
l
y
to
g
en
er
ate
d
i
v
er
s
e
y
et
r
ea
lis
tic
s
p
ee
ch
s
am
p
les.
Fo
r
ex
am
p
le,
a
f
ile
p
o
o
l
o
f
5
2
0
u
t
ter
an
ce
s
f
o
r
L
e
v
an
tin
e
r
eq
u
i
r
e
d
an
ad
d
itio
n
al
2
8
0
s
y
n
th
etica
lly
g
en
er
ated
s
am
p
l
es
,
wh
ich
is
3
5
%
o
f
th
e
g
r
o
u
p
’
s
d
ata,
to
ac
h
iev
e
th
e
tar
g
et
.
T
h
is
au
g
m
en
tatio
n
n
o
t
o
n
ly
b
alan
ce
d
g
r
o
u
p
s
izes
b
u
t
also
ex
p
an
d
e
d
ac
o
u
s
ti
c
v
ar
iab
ilit
y
,
ef
f
ec
tiv
ely
s
im
u
latin
g
s
p
ea
k
er
an
d
r
ec
o
r
d
in
g
d
i
v
er
s
ity
.
T
h
e
r
esu
l
t
was
4
,
0
0
0
p
r
o
ce
s
s
ed
f
iles
u
s
in
g
a
u
n
if
o
r
m
p
r
o
ce
d
u
r
e,
th
a
t
p
r
o
v
i
d
es
a
s
tr
o
n
g
b
asis
f
o
r
ac
ce
n
t
class
if
icatio
n
with
h
y
b
r
i
d
C
NN
-
B
iLST
M
-
b
ased
m
o
d
els.
T
h
e
r
esu
ltin
g
s
p
ec
tr
o
g
r
am
s
ar
e
s
h
o
wn
in
Fig
u
r
e
2
(
Sp
ec
tr
o
g
r
a
m
s
o
f
Au
g
m
e
n
ted
Au
d
i
o
Sig
n
als)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
4
,
Dec
em
b
er
20
26
:
1
6
7
3
-
1
6
8
1
1676
Fig
u
r
e
1
.
Dis
tr
ib
u
tio
n
o
f
th
e
f
i
n
al
Ar
ab
ic
s
p
ee
ch
d
ataset
am
o
n
g
f
iv
e
d
ialec
t g
r
o
u
p
s
: L
ev
an
ti
n
e,
Gu
lf
,
Nile
Valley
,
Me
s
o
p
o
tam
ian
,
a
n
d
No
r
th
Af
r
ican
Fig
u
r
e
2
.
Sp
ec
tr
o
g
r
am
s
o
f
au
g
m
en
ted
au
d
io
s
ig
n
als
3
.
3
.
M
o
del
a
rc
hite
ct
ure
T
h
e
p
r
o
p
o
s
ed
ar
c
h
itectu
r
e
in
c
lu
d
es
co
n
v
o
lu
tio
n
al
an
d
r
ec
u
r
r
en
t
lay
er
s
to
d
etec
t
lo
ca
l
a
n
d
s
eq
u
en
tial
p
atter
n
s
in
th
e
in
p
u
t
em
b
e
d
d
i
n
g
s
.
T
h
e
in
p
u
t
o
f
ea
ch
in
s
tan
ce
,
wh
ich
is
in
th
e
f
o
r
m
o
f
a
s
eq
u
en
ce
o
f
2
4
-
tim
e
s
tep
s
with
3
2
-
d
im
en
s
io
n
al
f
ea
tu
r
es,
b
ec
o
m
es
s
u
b
ject
to
p
r
o
c
ess
in
g
b
y
a
one
-
d
im
en
s
io
n
al
c
o
n
v
o
lu
ti
o
n
al
b
lo
c
k
in
th
e
f
o
r
m
o
f
a
b
u
ild
in
g
b
lo
c
k
.
T
h
is
co
n
v
o
lu
tio
n
al
b
lo
c
k
c
o
n
tain
s
a
co
n
v
o
lu
tio
n
lay
er
,
a
n
o
r
m
aliza
tio
n
la
y
er
,
an
ac
tiv
atio
n
f
u
n
ctio
n
,
an
d
a
p
o
o
lin
g
lay
er
,
wh
ich
to
g
eth
e
r
p
er
f
o
r
m
lo
ca
l
p
atter
n
d
is
co
v
er
y
an
d
r
e
d
u
ce
d
ata
len
g
th
wh
ile
im
p
r
o
v
in
g
co
m
p
u
tatio
n
al
ef
f
icien
cy
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
cc
en
t
-
a
w
a
r
e
a
r
a
b
ic
s
p
ee
ch
p
r
o
ce
s
s
in
g
u
s
in
g
C
N
N
–
B
i
LS
T
M fo
r
d
ia
lect
…
(
Ta
r
ik
A
b
d
el
-
K
a
r
ee
m
A
b
u
A
in
)
1677
T
h
e
r
es
u
ltin
g
f
ea
tu
r
e
m
ap
s
ar
e
in
p
u
t
in
to
a
b
id
ir
ec
tio
n
a
l
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
B
iLST
M)
n
etwo
r
k
th
at
ca
p
t
u
r
es
lo
n
g
-
r
a
n
g
e
d
ep
e
n
d
en
cies
in
b
o
th
d
ir
e
ctio
n
s
.
B
y
co
m
b
in
in
g
th
e
f
o
r
war
d
an
d
b
ac
k
war
d
h
id
d
en
s
tates,
th
e
m
o
d
el
cr
ea
t
es
a
5
1
2
-
d
im
en
s
io
n
al
r
ep
r
esen
tatio
n
o
f
th
e
s
eq
u
en
ce
.
T
h
is
r
e
p
r
esen
tatio
n
is
th
e
n
f
ed
in
to
a
f
u
lly
co
n
n
ec
ted
f
ee
d
f
o
r
war
d
n
etwo
r
k
with
R
eL
U
ac
tiv
atio
n
,
d
r
o
p
o
u
t
r
eg
u
lar
izatio
n
,
an
d
an
o
u
tp
u
t
-
d
im
en
s
io
n
m
a
p
p
in
g
to
its
f
in
al
s
p
ac
e.
T
o
r
ed
u
ce
o
v
er
f
itti
n
g
i
n
p
r
ed
i
ctio
n
s
an
d
im
p
r
o
v
e
g
en
e
r
aliza
tio
n
,
th
e
m
o
d
el
u
s
es
a
lab
el
s
m
o
o
th
in
g
cr
o
s
s
-
en
tr
o
p
y
l
o
s
s
.
T
h
e
tr
ain
i
n
g
p
r
o
ce
s
s
u
s
es
ad
ap
tiv
e
lear
n
in
g
-
r
ate
s
ch
e
d
u
lin
g
a
n
d
ea
r
l
y
s
to
p
p
in
g
t
o
en
s
u
r
e
r
o
b
u
s
t c
o
n
v
er
g
e
n
ce
.
All
in
all,
C
NN
-
B
iLST
M
in
t
eg
r
ates
lo
ca
l
f
ea
tu
r
e
ex
tr
ac
tio
n
with
a
s
eq
u
en
tial
m
o
d
elin
g
ap
p
r
o
ac
h
,
m
ak
in
g
it
co
m
p
ar
a
b
le
to
m
eth
o
d
s
th
at
r
ely
o
n
s
tr
u
ctu
r
e
d
r
ep
r
esen
tatio
n
s
an
d
s
eq
u
en
ce
-
b
ased
class
if
icatio
n
.
T
h
e
m
o
d
el
was
tr
ain
ed
u
s
in
g
Ad
am
(
lr
=
1
×1
0
⁻³)
,
b
atch
s
ize
3
2
,
u
p
to
5
0
ep
o
ch
s
with
ea
r
ly
s
to
p
p
in
g
(
p
atien
ce
=5
)
,
1
2
8
C
NN
ch
a
n
n
els,
2
5
6
B
iLST
M
h
id
d
en
u
n
i
ts
,
an
d
0
.
1
lab
el
s
m
o
o
t
h
in
g
.
T
h
e
o
v
er
all
ar
ch
itectu
r
e
is
illu
s
tr
ated
in
Fig
u
r
e
3
.
Fig
u
r
e
3
.
R
esear
ch
m
o
d
el
ar
ch
itectu
r
e
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
As
s
h
o
wn
in
Fig
u
r
e
4
,
th
e
c
o
n
f
u
s
io
n
m
atr
ix
s
h
o
ws
o
n
ly
a
f
ew
m
is
class
if
icatio
n
s
,
m
o
s
tly
am
o
n
g
lin
g
u
is
tically
s
im
ilar
d
ialec
t
s
,
in
d
icatin
g
th
e
m
o
d
el’
s
ab
ilit
y
to
d
etec
t
s
u
b
tle
lin
g
u
is
tic
d
if
f
er
en
ce
s
.
Mo
r
eo
v
er
,
it a
ls
o
s
h
o
ws th
at
L
ev
an
tin
e
an
d
Nile
Valley
d
i
alec
ts
ar
e
f
r
eq
u
en
tly
co
n
f
u
s
ed
with
th
e
Gu
lf
d
ialec
t
,
with
1
3
an
d
1
1
m
is
class
if
ied
s
am
p
les,
r
esp
ec
tiv
ely
,
p
r
o
b
ab
l
y
d
u
e
to
p
h
o
n
etic
s
im
ilar
ities
ac
r
o
s
s
n
eig
h
b
o
r
in
g
r
eg
io
n
s
.
Sin
ce
th
e
m
o
d
el
r
elies
o
n
s
p
ec
tr
o
g
r
am
s
r
ath
er
t
h
an
tex
tu
al
d
ata,
th
ese
er
r
o
r
s
m
ain
ly
s
tem
f
r
o
m
ac
o
u
s
tic
s
im
ilar
ity
.
I
n
co
r
p
o
r
at
in
g
p
h
o
n
em
e
-
le
v
el
f
ea
tu
r
es o
r
lex
ical
cu
es in
f
u
tu
r
e
iter
atio
n
s
co
u
ld
h
elp
r
e
d
u
ce
th
ese
d
ialec
t m
is
class
if
icatio
n
s
.
T
h
e
class
if
icatio
n
r
ep
o
r
t
T
a
b
l
e
3
co
n
f
ir
m
s
th
e
m
o
d
el’
s
r
o
b
u
s
tn
ess
,
ex
h
ib
itin
g
an
o
v
er
all
a
cc
u
r
ac
y
o
f
8
8
%.
B
o
th
m
ac
r
o
-
an
d
weig
h
ted
-
av
er
ag
e
d
m
etr
ics
f
o
r
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
Sco
r
e
ar
e
id
en
tical
at
0
.
8
8
.
T
h
e
Me
s
o
p
o
tam
ia
n
d
ialec
t
ac
h
iev
es
th
e
h
i
g
h
est
F1
-
s
co
r
e
o
f
0
.
9
2
,
r
ef
lectin
g
ex
ce
llen
t
p
r
ec
is
io
n
(
0
.
9
3
)
a
n
d
r
ec
all
(
0
.
9
1
)
.
Fu
r
th
er
m
o
r
e,
th
e
Gu
lf
an
d
No
r
th
Af
r
ica
d
ialec
t
s
p
er
f
o
r
m
ed
s
tr
o
n
g
ly
,
with
F1
-
s
co
r
es
o
f
0
.
8
5
a
n
d
0
.
8
8
,
r
esp
ec
tiv
ely
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
4
,
Dec
em
b
er
20
26
:
1
6
7
3
-
1
6
8
1
1678
Fig
u
r
e
4
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
th
e
d
ialec
t c
lass
if
icatio
n
r
esu
lts
T
ab
le
3
.
C
lass
if
icatio
n
r
ep
o
r
t
s
h
o
win
g
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
es b
y
d
ialec
t
C
l
a
s
ses
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
S
u
p
p
o
r
t
G
u
l
f
0
.
8
1
0
.
9
0
0
.
8
5
1
5
9
Le
v
a
n
t
i
n
e
0
.
9
1
0
.
8
2
0
.
8
6
1
6
6
M
e
s
o
p
o
t
a
m
-
i
a
n
0
.
9
3
0
.
9
1
0
.
9
2
1
6
4
N
i
l
e
_
V
a
l
l
e
y
0
.
9
1
0
.
8
5
0
.
8
8
1
7
3
N
o
r
t
h
_
A
f
r
i
c
a
0
.
8
4
0
.
9
3
0
.
8
8
1
3
8
T
h
o
u
g
h
L
ev
a
n
tin
e
r
ec
o
r
d
ed
a
s
lig
h
tly
lo
wer
r
ec
all
(
0
.
8
2
)
,
it
m
ain
tain
ed
h
ig
h
p
r
ec
is
io
n
(
0
.
9
1
)
,
in
d
icatin
g
th
at
p
r
ed
ictio
n
s
f
o
r
th
is
class
wer
e
co
n
s
er
v
ativ
e
y
et
r
eliab
le.
T
h
e
Nile
Valley
d
ialec
t
s
h
o
wed
co
n
s
is
ten
t
r
esu
lts
with
an
F1
-
s
co
r
e
o
f
0
.
8
8
,
f
u
r
th
er
c
o
n
f
ir
m
in
g
b
alan
ce
d
p
er
f
o
r
m
an
ce
ac
r
o
s
s
class
e
s
.
T
h
e
lear
n
i
n
g
c
u
r
v
es
i
n
Fig
u
r
e
5
s
h
o
w
s
m
o
o
th
co
n
v
er
g
en
ce
,
with
ac
cu
r
ac
y
s
tead
ily
im
p
r
o
v
in
g
a
n
d
lo
s
s
d
ec
lin
in
g
th
r
o
u
g
h
o
u
t
tr
ai
n
in
g
.
T
h
e
clo
s
e
alig
n
m
e
n
t
b
et
wee
n
tr
ain
in
g
a
n
d
v
a
lid
atio
n
t
r
ajec
to
r
ies
in
d
icate
s
th
e
ab
s
en
ce
o
f
o
v
er
f
itti
n
g
,
th
er
eb
y
v
alid
atin
g
th
e
s
tab
ilit
y
,
r
eliab
ilit
y
,
an
d
g
en
er
aliza
b
ilit
y
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el.
Fig
u
r
e
5
.
T
r
ain
in
g
a
n
d
v
alid
atio
n
lo
s
s
an
d
ac
cu
r
ac
y
g
r
ap
h
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
cc
en
t
-
a
w
a
r
e
a
r
a
b
ic
s
p
ee
ch
p
r
o
ce
s
s
in
g
u
s
in
g
C
N
N
–
B
i
LS
T
M fo
r
d
ia
lect
…
(
Ta
r
ik
A
b
d
el
-
K
a
r
ee
m
A
b
u
A
in
)
1679
I
n
iti
all
y
,
t
h
e
s
y
s
te
m
atic
a
u
g
m
en
t
ati
o
n
m
et
h
o
d
ta
ck
le
d
t
h
e
p
r
o
b
l
em
o
f
u
n
e
v
e
n
c
lass
d
is
t
r
i
b
u
ti
o
n
a
n
d
in
t
r
o
d
u
ce
d
ac
o
u
s
ti
c
d
i
v
e
r
s
it
y
b
y
s
i
m
u
la
ti
n
g
t
h
e
p
h
y
s
i
ca
l
d
i
f
f
e
r
e
n
c
es
b
e
twe
en
s
p
ea
k
er
s
.
T
h
is
e
n
a
b
le
d
t
h
e
m
o
d
el
to
g
e
n
e
r
a
liz
e
m
o
r
e
ef
f
e
cti
v
e
ly
a
cr
o
s
s
d
i
ale
cts
r
at
h
e
r
t
h
a
n
o
v
er
f
it
ti
n
g
to
t
h
e
m
o
s
t
f
r
e
q
u
e
n
t
class
es.
M
o
r
e
o
v
e
r
,
th
e
h
y
b
r
i
d
C
NN
-
B
iLS
T
M
m
o
d
el
was
f
o
u
n
d
to
b
e
a
d
v
a
n
t
ag
eo
u
s
s
i
n
c
e
i
t
c
o
m
b
i
n
e
d
t
h
e
e
x
t
r
a
cti
o
n
o
f
lo
ca
l
s
p
e
ct
r
al
f
e
at
u
r
es
wi
th
t
h
e
c
a
p
a
cit
y
t
o
c
a
p
t
u
r
e
l
o
n
g
-
r
a
n
g
e
d
e
p
e
n
d
e
n
c
ies
b
e
twe
en
c
o
n
s
ec
u
ti
v
e
o
b
s
e
r
v
a
ti
o
n
s
,
a
f
e
at
u
r
e
w
h
i
c
h
is
i
m
p
o
r
t
an
t
f
o
r
d
is
ti
n
g
u
is
h
i
n
g
s
u
b
tl
e
v
o
c
al
d
i
f
f
er
e
n
ce
s
i
n
cl
o
s
e
ly
r
el
at
ed
d
ia
lects
.
T
h
e
m
ain
r
ea
s
o
n
f
o
r
th
e
m
is
class
if
icatio
n
s
in
th
e
co
n
f
u
s
io
n
m
atr
ix
is
th
e
p
r
esen
ce
o
f
s
im
ilar
ter
m
s
,
wh
ich
s
h
o
ws
th
at
th
e
task
its
elf
is
d
if
f
icu
lt
,
r
ath
er
th
an
p
o
in
tin
g
to
an
y
s
h
o
r
tco
m
i
n
g
s
in
th
e
m
eth
o
d
o
lo
g
y
.
Mo
r
eo
v
er
,
th
e
s
tab
ilit
y
o
f
th
e
lear
n
in
g
cu
r
v
es
co
n
f
i
r
m
s
th
e
r
o
b
u
s
tn
ess
o
f
th
e
o
p
tim
izatio
n
p
r
o
ce
s
s
an
d
s
h
o
ws
th
at
th
er
e
is
n
o
o
v
er
f
itti
n
g
.
W
h
en
co
m
p
a
r
ed
with
p
r
e
v
io
u
s
s
tu
d
ies
th
at
u
s
ed
s
h
a
llo
w
class
if
ier
s
o
r
s
in
g
le
-
s
tr
ea
m
n
eu
r
al
n
etwo
r
k
s
,
th
e
p
r
esen
t
ex
p
er
im
e
n
t
ac
h
iev
es
a
b
etter
b
alan
ce
b
etw
ee
n
ac
cu
r
ac
y
an
d
g
en
er
aliza
tio
n
.
T
h
is
s
u
g
g
ests
th
at
C
NN
-
B
iLST
M
m
o
d
els
h
av
e
p
r
o
m
is
e
as
a
v
alu
a
b
l
e
way
f
o
r
f
u
tu
r
e
d
ev
elo
p
m
e
n
t
in
ac
ce
n
t
-
s
en
s
itiv
e
Ar
a
b
ic
s
p
ee
ch
tech
n
o
l
o
g
ie
s
.
T
ab
le
4
d
em
o
n
s
tr
ates
a
co
m
p
ar
ativ
e
s
u
m
m
ar
y
o
f
o
u
r
m
eth
o
d
in
c
o
m
p
ar
is
o
n
t
o
ex
is
tin
g
ap
p
r
o
ac
h
es
th
at
also
u
s
ed
s
im
ilar
Yo
u
T
u
b
e
-
b
ased
d
atasets
.
T
ab
le
4
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
with
ex
is
tin
g
m
eth
o
d
s
S
t
u
d
y
M
o
d
e
l
D
a
t
a
s
e
t
A
c
c
u
r
a
c
y
(
%)
A
l
sh
a
b
a
n
a
h
a
n
d
A
n
n
a
v
a
r
a
m
[
1
]
A
d
e
e
p
l
e
a
r
n
i
n
g
c
l
a
ssi
f
i
c
a
t
i
o
n
mo
d
e
l
O
v
e
r
6
6
,
0
0
0
a
u
d
i
o
f
i
l
e
s
f
r
o
m
t
h
e
Y
o
u
Tu
b
e
ADI
-
1
7
M
I
T
d
a
t
a
se
t
A
c
h
i
e
v
e
d
8
6
%
a
c
c
u
r
a
c
y
i
n
t
e
s
t
d
a
t
a
K
e
l
e
g
e
t
a
l
.
[
2
]
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
m
o
d
e
l
s
(
K
-
N
e
a
r
e
st
N
e
i
g
h
b
o
r
(
K
N
N
)
,
R
a
n
d
o
m F
o
r
e
s
t
(
R
F
)
,
M
u
l
t
i
-
La
y
e
r
P
e
r
c
e
p
t
r
o
n
(
M
LP)
,
a
n
d
A
r
t
i
f
i
c
i
a
l
N
e
u
r
a
l
N
e
t
w
o
r
k
s
(ANN))
Th
e
d
a
t
a
se
t
w
a
s
g
a
t
h
e
r
e
d
f
r
o
m Y
o
u
T
u
b
e
v
i
d
e
o
s
a
n
d
a
u
d
i
o
s
e
g
me
n
t
s
l
a
b
e
l
e
d
b
y
h
u
m
a
n
e
x
p
e
r
t
s
-
M
F
C
C
f
e
a
t
u
r
e
s
K
N
N
:
6
2
%,
R
F
:
6
0
%
M
LP:
3
3
%,
A
N
N
:
4
2
%
-
TFCC
F
e
a
t
u
r
e
s
K
N
N
:
7
6
%,
R
F
:
6
4
%
M
LP:
3
4
%,
A
N
N
:
4
1
%
Th
e
p
r
o
p
o
s
e
d
met
h
o
d
H
y
b
r
i
d
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
-
b
i
d
i
r
e
c
t
i
o
n
a
l
l
o
n
g
sh
o
r
t
-
t
e
r
m
mem
o
r
y
(
C
N
N
-
B
i
LST
M
)
m
o
d
e
l
Th
e
d
a
t
a
se
t
w
a
s
c
o
l
l
e
c
t
e
d
f
r
o
m
v
a
r
i
o
u
s
Y
o
u
T
u
b
e
v
i
d
e
o
s
a
n
d
d
i
v
i
d
e
d
i
n
t
o
f
i
v
e
c
a
t
e
g
o
r
y
g
r
o
u
p
s
,
s
u
c
h
a
s L
e
v
a
n
t
i
n
e
,
G
u
l
f
,
N
i
l
e
V
a
l
l
e
y
,
M
e
s
o
p
o
t
a
m
i
a
n
,
a
n
d
N
o
r
t
h
A
f
r
i
c
a
n
A
c
h
i
e
v
e
d
8
8
%
a
c
c
u
r
a
c
y
i
n
t
e
s
t
d
a
t
a
5.
CO
NCLU
SI
O
N
T
h
e
p
ap
er
s
h
o
ws
th
at
th
e
C
NN
-
B
iLST
M
f
r
am
ewo
r
k
d
ea
ls
with
th
e
lo
n
g
-
s
tan
d
in
g
p
r
o
b
lem
o
f
class
if
y
in
g
Ar
ab
ic
d
ialec
ts
,
ac
h
iev
in
g
an
o
v
er
all
ac
cu
r
ac
y
o
f
8
8
% a
n
d
s
h
o
win
g
co
n
s
is
ten
t p
er
f
o
r
m
an
ce
with
in
f
iv
e
d
ialec
t
g
r
o
u
p
s
.
W
h
at
m
ak
es
th
e
m
o
d
el
s
p
ec
ial
is
its
h
y
b
r
id
a
r
ch
itectu
r
e,
th
is
ar
ch
itectu
r
e
co
m
b
in
in
g
co
n
v
o
l
u
tio
n
al
lay
e
r
s
th
at
p
ick
o
u
t
lo
ca
l
s
p
ec
tr
al
f
ea
tu
r
es
wi
th
B
iLST
M
u
n
its
ca
p
ab
le
o
f
ca
p
tu
r
in
g
tem
p
o
r
al
d
ep
en
d
e
n
cies.
I
t
is
b
ec
au
s
e
o
f
th
is
d
esig
n
th
at
th
e
m
o
d
el
i
s
ab
le
to
ef
f
ec
tiv
ely
tell
th
e
d
if
f
er
en
ce
b
etwe
en
clo
s
ely
r
elate
d
d
ialec
ts
.
Mo
r
e
o
v
er
,
th
e
s
tr
u
ctu
r
ed
d
ata
au
g
m
en
tatio
n
s
tr
ateg
y
c
o
m
b
ated
th
e
p
r
o
b
lem
o
f
class
im
b
alan
ce
b
y
in
t
r
o
d
u
cin
g
ac
o
u
s
tic
v
ar
iab
ilit
y
,
th
u
s
im
p
r
o
v
i
n
g
r
o
b
u
s
tn
ess
in
r
ea
l
-
wo
r
ld
s
itu
atio
n
s
.
E
v
en
th
o
u
g
h
th
e
d
ataset
h
as
b
ee
n
b
alan
ce
d
u
s
in
g
au
g
m
en
t
atio
n
,
th
e
Yo
u
T
u
b
e
r
ec
o
r
d
in
g
s
in
clu
d
e
u
n
co
n
tr
o
lled
n
o
is
e
an
d
v
ar
iati
o
n
s
b
etwe
en
c
h
an
n
els
,
w
h
ich
im
p
ed
e
th
e
m
o
d
el'
s
ab
i
lity
to
g
en
er
alize
to
clea
r
s
p
ee
ch
.
T
h
e
d
is
tin
ctio
n
b
etwe
en
d
ialec
t
g
r
o
u
p
s
s
u
ch
as
I
r
aq
i
an
d
Ku
waiti
o
r
b
etwe
en
Alg
e
r
ian
an
d
Mo
r
o
cc
a
n
s
till
r
em
ain
s
u
n
r
eso
lv
ed
,
m
ak
in
g
it
d
if
f
icu
lt
to
a
p
p
ly
th
e
m
o
d
el
to
m
in
o
r
d
ialec
t
d
if
f
er
en
ce
s
.
Mo
r
eo
v
er
,
th
e
tr
ain
-
test
s
p
lit
was
ca
r
r
ied
o
u
t
at
th
e
s
am
p
le
l
ev
el,
wh
ich
m
ig
h
t
r
esu
lt
in
ex
ce
s
s
iv
ely
o
p
tim
is
ti
c
e
s
tim
ates
o
f
ac
cu
r
ac
y
.
Fu
tu
r
e
wo
r
k
m
ig
h
t
s
tu
d
y
tr
a
n
s
f
o
r
m
er
-
b
ased
au
d
io
m
o
d
els
s
u
ch
as
W
h
is
p
er
an
d
W
av
2
Vec
,
as
well
a
s
x
-
v
ec
to
r
em
b
ed
d
in
g
s
,
in
o
r
d
er
to
in
cr
ea
s
e
s
p
ea
k
er
r
o
b
u
s
tn
ess
.
Mo
r
eo
v
er
,
th
e
d
ev
elo
p
m
en
t
o
f
m
u
ltimo
d
al
ASR
s
y
s
tem
s
th
at
in
teg
r
ate
b
o
th
ac
o
u
s
tic
an
d
tex
tu
a
l
cu
es
co
u
ld
also
co
n
tr
ib
u
te
to
b
etter
d
ialec
t c
o
v
er
ag
e.
ACK
NO
WL
E
DG
M
E
N
T
S
T
h
e
r
esear
ch
wo
r
k
was
s
elf
-
s
u
p
p
o
r
ted
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
e
r
esear
ch
wo
r
k
was
s
elf
-
f
u
n
d
ed
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
4
,
Dec
em
b
er
20
26
:
1
6
7
3
-
1
6
8
1
1680
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
F
o
I
R
D
O
E
Vi
Su
P
Fu
T
ar
ik
Ab
d
el
-
Kar
ee
m
Ab
u
Ain
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
W
alee
d
Ab
d
el
Kar
im
Ab
u
ain
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
T
h
e
au
th
o
r
s
tates n
o
co
n
f
lict o
f
in
ter
est.
I
NF
O
RM
E
D
CO
NS
E
N
T
No
t
Ap
p
licab
le
–
th
is
r
esear
ch
r
eq
u
ir
es n
o
h
u
m
an
o
r
in
d
i
v
id
u
al
co
n
s
en
t.
E
T
H
I
CAL AP
P
RO
V
AL
No
t
Ap
p
licab
le
–
th
is
r
esear
ch
d
id
n
o
t
in
v
o
l
v
e
an
y
h
u
m
a
n
p
a
r
ticip
atio
n
.
DATA AV
AI
L
AB
I
L
I
T
Y
Data
s
ets we
r
e
co
llected
f
r
o
m
v
ar
io
u
s
Yo
u
T
u
b
e
v
id
eo
s
.
RE
F
E
R
E
NC
E
S
[
1
]
A
.
A
l
sh
a
b
a
n
a
h
a
n
d
M
.
A
n
n
a
v
a
r
a
m,
“
O
n
u
si
n
g
A
r
a
b
i
c
l
a
n
g
u
a
g
e
d
i
a
l
e
c
t
s
i
n
r
e
c
o
mm
e
n
d
a
t
i
o
n
sy
s
t
e
ms
,
”
i
n
Fi
n
d
i
n
g
s
o
f
t
h
e
Asso
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
st
i
c
s:
N
A
AC
L
2
0
2
5
,
2
0
2
5
,
p
p
.
2
1
7
8
–
2
1
8
6
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
5
.
f
i
n
d
i
n
g
s
-
n
a
a
c
l
.
1
1
5
.
[
2
]
A
.
K
e
l
e
g
,
S
.
G
o
l
d
w
a
t
e
r
,
a
n
d
W
.
M
a
g
d
y
,
“
R
e
v
i
si
t
i
n
g
c
o
mm
o
n
a
ss
u
m
p
t
i
o
n
s
a
b
o
u
t
A
r
a
b
i
c
D
i
a
l
e
c
t
s
i
n
N
LP,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
6
3
r
d
An
n
u
a
l
M
e
e
t
i
n
g
o
f
t
h
e
Ass
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
st
i
c
s
(
Vo
l
u
m
e
1
:
L
o
n
g
Pa
p
e
rs)
,
2
0
2
5
,
p
p
.
3
3
0
9
–
3
3
2
7
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
5
.
a
c
l
-
l
o
n
g
.
1
6
6
.
[
3
]
M
.
H
a
b
b
a
s
h
e
t
a
l
.
,
“
R
e
c
o
g
n
i
t
i
o
n
o
f
A
r
a
b
i
c
a
c
c
e
n
t
s
f
r
o
m
En
g
l
i
s
h
s
p
o
k
e
n
s
p
e
e
c
h
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
,
”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
3
7
2
1
9
–
3
7
2
3
0
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
a
c
c
e
s
s.
2
0
2
4
.
3
3
7
4
7
6
8
.
[
4
]
K
.
K
a
sh
i
f
,
A
.
A
l
w
a
n
,
Y
.
W
u
,
L
.
D
e
N
a
r
d
i
s
,
a
n
d
M
.
-
G
.
D
i
B
e
n
e
d
e
t
t
o
,
“
M
K
EL
M
b
a
s
e
d
mu
l
t
i
-
c
l
a
ssi
f
i
c
a
t
i
o
n
mo
d
e
l
f
o
r
f
o
r
e
i
g
n
a
c
c
e
n
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
,
”
H
e
l
i
y
o
n
,
v
o
l
.
1
0
,
n
o
.
1
6
,
p
.
e
3
6
4
6
0
,
A
u
g
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
h
e
l
i
y
o
n
.
2
0
2
4
.
e
3
6
4
6
0
.
[
5
]
A
.
A
l
i
w
y
,
H
.
Ta
h
e
r
,
a
n
d
Z.
A
b
o
A
l
t
a
h
e
e
n
,
“
A
r
a
b
i
c
d
i
a
l
e
c
t
s
i
d
e
n
t
i
f
i
c
a
t
i
o
n
f
o
r
a
l
l
A
r
a
b
i
c
c
o
u
n
t
r
i
e
s,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
Fi
f
t
h
Ar
a
b
i
c
N
a
t
u
r
a
l
L
a
n
g
u
a
g
e
Pro
c
e
ss
i
n
g
W
o
rks
h
o
p
,
p
p
.
3
0
2
–
3
0
7
,
2
0
2
0
.
[
6
]
M
.
El
m
a
h
d
y
,
R
.
G
r
u
h
n
,
a
n
d
W
.
M
i
n
k
e
r
,
N
o
v
e
l
t
e
c
h
n
i
q
u
e
s f
o
r
d
i
a
l
e
c
t
a
l
Ara
b
i
c
s
p
e
e
c
h
re
c
o
g
n
i
t
i
o
n
.
S
p
r
i
n
g
e
r
U
S
,
2
0
1
2
.
[
7
]
N
.
A
l
sa
d
h
a
n
,
“
A
n
o
v
e
l
d
i
a
l
e
c
t
-
a
w
a
r
e
f
r
a
mew
o
r
k
f
o
r
t
h
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
A
r
a
b
i
c
D
i
a
l
e
c
t
s
a
n
d
e
m
o
t
i
o
n
s
,
”
J
o
u
r
n
a
l
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
2
1
,
n
o
.
1
,
p
p
.
8
8
–
9
5
,
Ja
n
.
2
0
2
5
,
d
o
i
:
1
0
.
3
8
4
4
/
j
c
ss
p
.
2
0
2
5
.
8
8
.
9
5
.
[
8
]
W
.
M
.
S
.
Y
a
f
o
o
z
,
“
E
n
h
a
n
c
i
n
g
A
r
a
b
i
c
d
i
a
l
e
c
t
d
e
t
e
c
t
i
o
n
o
n
s
o
c
i
a
l
m
e
d
i
a
:
a
h
y
b
r
i
d
m
o
d
e
l
w
i
t
h
a
n
a
t
t
e
n
t
i
o
n
me
c
h
a
n
i
sm,
”
I
n
f
o
rm
a
t
i
o
n
(
S
w
i
t
zer
l
a
n
d
)
,
v
o
l
.
1
5
,
n
o
.
6
,
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
i
n
f
o
1
5
0
6
0
3
1
6
.
[
9
]
H
.
El
saf
t
y
,
B
.
A
b
d
o
u
,
T.
D
e
u
ß
e
r
,
M
.
P
i
e
l
k
a
,
C
.
B
a
u
c
k
h
a
g
e
,
a
n
d
R
.
S
i
f
a
,
“
A
r
D
i
a
:
i
m
p
r
o
v
i
n
g
a
r
a
b
i
c
d
i
a
l
e
c
t
a
l
l
a
n
g
u
a
g
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
a
n
o
v
e
l
d
a
t
a
se
t
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
n
t
e
rn
a
t
i
o
n
a
l
A
AAI
C
o
n
f
e
re
n
c
e
o
n
W
e
b
a
n
d
S
o
c
i
a
l
Me
d
i
a
,
v
o
l
.
1
9
,
n
o
.
I
c
w
sm,
p
p
.
2
4
1
3
–
2
4
2
2
,
2
0
2
5
,
d
o
i
:
1
0
.
1
6
0
9
/
i
c
w
sm.
v
1
9
i
1
.
3
5
9
4
4
.
[
1
0
]
H
.
S
a
l
e
h
e
t
a
l
.
,
“
A
d
v
a
n
c
i
n
g
a
r
a
b
i
c
d
i
a
l
e
c
t
d
e
t
e
c
t
i
o
n
w
i
t
h
h
y
b
r
i
d
st
a
c
k
e
d
t
r
a
n
sf
o
r
mer
mo
d
e
l
s,
”
F
ro
n
t
i
e
rs
i
n
H
u
m
a
n
N
e
u
ro
s
c
i
e
n
c
e
,
v
o
l
.
1
9
,
2
0
2
5
,
d
o
i
:
1
0
.
3
3
8
9
/
f
n
h
u
m.
2
0
2
5
.
1
4
9
8
2
9
7
.
[
1
1
]
B
.
M
.
A
b
d
u
l
l
a
h
,
M
.
B
a
a
s,
B
.
M
ö
b
i
u
s,
a
n
d
D
.
K
l
a
k
o
w
,
“
V
o
i
c
e
c
o
n
v
e
r
si
o
n
i
m
p
r
o
v
e
s
c
r
o
ss
-
d
o
ma
i
n
r
o
b
u
s
t
n
e
ss
f
o
r
sp
o
k
e
n
a
r
a
b
i
c
d
i
a
l
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
,
”
2
0
2
5
.
[
1
2
]
R
.
A
l
h
a
r
b
i
,
“
M
F
-
S
a
u
d
i
:
A
m
u
l
t
i
mo
d
a
l
f
r
a
mew
o
r
k
f
o
r
b
r
i
d
g
i
n
g
t
h
e
g
a
p
b
e
t
w
e
e
n
a
u
d
i
o
a
n
d
t
e
x
t
u
a
l
d
a
t
a
f
o
r
S
a
u
d
i
d
i
a
l
e
c
t
d
e
t
e
c
t
i
o
n
,
”
J
o
u
rn
a
l
o
f
K
i
n
g
S
a
u
d
U
n
i
v
e
rsi
t
y
-
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
c
i
e
n
c
e
s
,
v
o
l
.
3
6
,
n
o
.
6
,
p
.
1
0
2
0
8
4
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
k
s
u
c
i
.
2
0
2
4
.
1
0
2
0
8
4
.
[
1
3
]
A
.
A
.
A
l
su
w
a
y
l
i
m
i
,
“
A
r
a
b
i
c
d
i
a
l
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
i
n
s
o
c
i
a
l
me
d
i
a
:
A
h
y
b
r
i
d
m
o
d
e
l
w
i
t
h
t
r
a
n
sf
o
r
m
e
r
mo
d
e
l
s
a
n
d
B
i
LST
M
,
”
H
e
l
i
y
o
n
,
v
o
l
.
1
0
,
n
o
.
1
7
,
p
.
e
3
6
2
8
0
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
h
e
l
i
y
o
n
.
2
0
2
4
.
e
3
6
2
8
0
.
[
1
4
]
N
.
Ti
b
i
a
n
d
M
.
A
n
o
u
a
r
B
e
n
M
e
ss
a
o
u
d
,
“
A
r
a
b
i
c
d
i
a
l
e
c
t
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
a
n
a
d
a
p
t
i
v
e
d
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
,
”
Bu
l
l
e
t
i
n
o
f
El
e
c
t
ri
c
a
l
E
n
g
i
n
e
e
ri
n
g
a
n
d
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
1
4
,
n
o
.
2
,
p
p
.
1
1
0
8
–
1
1
1
6
,
A
p
r
.
2
0
2
5
,
d
o
i
:
1
0
.
1
1
5
9
1
/
e
e
i
.
v
1
4
i
2
.
8
1
6
5
.
[
1
5
]
A
.
R
.
A
l
i
,
“
M
u
l
t
i
-
D
i
a
l
e
c
t
A
r
a
b
i
c
sp
e
e
c
h
r
e
c
o
g
n
i
t
i
o
n
,
”
i
n
2
0
2
0
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
i
n
t
C
o
n
f
e
r
e
n
c
e
o
n
N
e
u
r
a
l
N
e
t
w
o
rk
s (I
J
C
N
N
)
,
2
0
2
0
,
p
p
.
1
–
7
,
d
o
i
:
1
0
.
1
1
0
9
/
i
j
c
n
n
4
8
6
0
5
.
2
0
2
0
.
9
2
0
6
6
5
8
.
[
1
6
]
M
.
A
l
-
A
y
y
o
u
b
,
M
.
K
.
R
i
h
a
n
i
,
N
.
I
.
D
a
l
g
a
m
o
n
i
,
a
n
d
N
.
A
.
A
b
d
u
l
l
a
,
“
S
p
o
k
e
n
A
r
a
b
i
c
d
i
a
l
e
c
t
s
i
d
e
n
t
i
f
i
c
a
t
i
o
n
:
Th
e
c
a
se
o
f
E
g
y
p
t
i
a
n
a
n
d
J
o
r
d
a
n
i
a
n
d
i
a
l
e
c
t
s,
”
i
n
2
0
1
4
5
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
I
n
f
o
rm
a
t
i
o
n
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
S
y
st
e
m
s
(
I
C
I
C
S
)
,
A
p
r
.
2
0
1
4
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
i
a
c
s
.
2
0
1
4
.
6
8
4
1
9
7
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
cc
en
t
-
a
w
a
r
e
a
r
a
b
ic
s
p
ee
ch
p
r
o
ce
s
s
in
g
u
s
in
g
C
N
N
–
B
i
LS
T
M fo
r
d
ia
lect
…
(
Ta
r
ik
A
b
d
el
-
K
a
r
ee
m
A
b
u
A
in
)
1681
[
1
7
]
F
.
S
.
A
l
–
A
n
z
i
a
n
d
B
.
S
.
S
.
Th
a
n
k
a
l
e
e
l
a
,
“
R
e
g
i
o
n
-
w
i
s
e
r
e
c
o
g
n
i
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
A
r
a
b
i
c
d
i
a
l
e
c
t
s
a
n
d
v
o
c
a
b
u
l
a
r
y
:
a
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
,
”
A
p
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
5
,
n
o
.
1
2
,
p
.
6
5
1
6
,
2
0
2
5
,
d
o
i
:
1
0
.
3
3
9
0
/
a
p
p
1
5
1
2
6
5
1
6
.
[
1
8
]
A
.
A
l
sa
h
e
e
l
,
N
.
A
l
m
a
t
r
a
f
i
,
N
.
A
l
h
u
m
a
i
d
,
D
.
M
.
I
b
r
a
h
i
m
,
a
n
d
A
.
A
.
A
l
-
S
h
a
r
g
a
b
i
,
“
E
n
h
a
n
c
i
n
g
A
r
a
b
i
c
d
i
a
l
e
c
t
c
l
a
ssi
f
i
c
a
t
i
o
n
w
i
t
h
d
e
e
p
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
,
”
i
n
2
0
2
5
5
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
Em
e
r
g
i
n
g
S
m
a
r
t
T
e
c
h
n
o
l
o
g
i
e
s
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
(
e
S
m
a
rT
A)
,
A
u
g
.
2
0
2
5
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
e
sm
a
r
t
a
6
6
7
6
4
.
2
0
2
5
.
1
1
1
3
2
2
7
3
.
[
1
9
]
G
.
A
l
-
S
h
w
a
y
y
a
t
a
n
d
O
.
N
.
G
e
r
e
k
,
“
H
y
b
r
i
d
d
e
e
p
l
e
a
r
n
i
n
g
a
n
d
s
i
g
n
a
l
p
r
o
c
e
ssi
n
g
f
o
r
A
r
a
b
i
c
d
i
a
l
e
c
t
r
e
c
o
g
n
i
t
i
o
n
i
n
l
o
w
-
r
e
so
u
r
c
e
set
t
i
n
g
s
,
”
a
rX
i
v
p
r
e
p
r
i
n
t
a
rX
i
v
:
2
5
0
6
.
2
1
3
8
6
,
2
0
2
5
,
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s/
2
5
0
6
.
2
1
3
8
6
.
[
2
0
]
B
.
T
a
l
a
f
h
a
e
t
a
l
.
,
“
N
A
D
I
2
0
2
5
:
t
h
e
f
i
r
st
m
u
l
t
i
d
i
a
l
e
c
t
a
l
A
r
a
b
i
c
sp
e
e
c
h
p
r
o
c
e
s
si
n
g
s
h
a
r
e
d
t
a
s
k
,
”
2
0
2
5
.
[
2
1
]
A
.
B
a
r
a
k
a
t
,
O
.
A
l
H
a
mm
a
d
i
,
A
.
A
l
d
h
a
h
e
r
i
,
a
n
d
A
.
E
l
n
a
g
a
r
,
“
A
r
a
b
i
c
d
i
a
l
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
f
r
o
m
sp
e
e
c
h
,
”
i
n
2
0
2
4
1
5
t
h
A
n
n
u
a
l
U
n
d
e
rg
r
a
d
u
a
t
e
Re
se
a
r
c
h
C
o
n
f
e
r
e
n
c
e
o
n
A
p
p
l
i
e
d
C
o
m
p
u
t
i
n
g
(
U
RC
)
,
A
p
r
.
2
0
2
4
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
u
r
c
6
2
2
7
6
.
2
0
2
4
.
1
0
6
0
4
5
5
7
.
[
2
2
]
K
.
M
.
O
.
N
a
h
a
r
,
O
.
M
.
A
l
-
H
a
z
a
i
me
h
,
A
.
A
b
u
-
E
i
n
,
a
n
d
M
.
A
.
A
l
-
B
e
t
a
r
,
“
A
r
a
b
i
c
d
i
a
l
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
u
s
i
n
g
d
i
f
f
e
r
e
n
t
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
me
t
h
o
d
s,
”
R
e
se
a
rc
h
S
q
u
a
r
e
,
p
p
.
1
–
2
1
,
2
0
2
2
.
[
2
3
]
M
.
H
a
m
i
d
i
e
t
a
l
.
,
“
M
i
x
e
d
A
S
R
sy
st
e
m
f
o
r
a
ma
z
i
g
h
a
n
d
A
r
a
b
i
c
u
n
d
e
r
r
e
so
u
r
c
e
d
d
i
a
l
e
c
t
s
i
n
m
a
g
h
r
e
b
r
e
g
i
o
n
,
”
2
0
2
4
,
d
o
i
:
1
0
.
3
1
2
1
9
/
o
sf
.
i
o
/
9
sw
5
4
.
[
2
4
]
S
.
A
l
-
A
z
a
n
i
a
n
d
E.
-
S
.
M
.
El
-
A
l
f
y
,
“
A
u
d
i
o
-
t
e
x
t
u
a
l
a
r
a
b
i
c
d
i
a
l
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
f
o
r
o
p
i
n
i
o
n
m
i
n
i
n
g
v
i
d
e
o
s
,
”
i
n
2
0
1
9
I
EEE
S
y
m
p
o
s
i
u
m
S
e
ri
e
s
o
n
C
o
m
p
u
t
a
t
i
o
n
a
l
I
n
t
e
l
l
i
g
e
n
c
e
(
S
S
C
I
)
,
D
e
c
.
2
0
1
9
,
p
p
.
2
4
7
0
–
2
4
7
5
,
d
o
i
:
1
0
.
1
1
0
9
/
ssc
i
4
4
8
1
7
.
2
0
1
9
.
9
0
0
3
0
3
1
.
[
2
5
]
S
.
M
a
l
m
a
si
a
n
d
M
.
Z
a
mp
i
e
r
i
,
“
A
r
a
b
i
c
d
i
a
l
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
i
n
sp
e
e
c
h
t
r
a
n
scri
p
t
s
,
"
I
n
p
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
t
h
i
rd
w
o
r
k
sh
o
p
o
n
N
L
P
f
o
r
si
m
i
l
a
r l
a
n
g
u
a
g
e
s,
v
a
ri
e
t
i
e
s
a
n
d
d
i
a
l
e
c
t
s
,
n
o
.
2
0
1
0
,
p
p
.
1
0
6
–
1
1
3
,
2
0
1
6
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Ta
r
ik
Ab
d
e
l
-
K
a
r
e
e
m
Abu
Ai
n
e
a
rn
e
d
h
is
first
d
e
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
fro
m
Ya
rm
o
u
k
Un
iv
e
rsit
y
in
Jo
r
d
a
n
i
n
2
0
0
3
.
He
we
n
t
o
n
t
o
o
b
tain
a
m
a
ste
r’s
d
e
g
re
e
in
c
o
m
p
u
ter
sc
ien
c
e
fro
m
AMU
Un
iv
e
rsity
,
a
lso
in
Jo
r
d
a
n
,
i
n
2
0
0
7
.
I
n
2
0
1
6
,
h
e
re
c
e
iv
e
d
h
is
P
h
D
i
n
Artifi
c
ial
In
telli
g
e
n
c
e
fro
m
th
e
A
rti
ficia
l
In
tell
ig
e
n
c
e
Tec
h
n
o
l
o
g
y
Ce
n
ter
a
t
th
e
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter
S
c
ien
c
e
,
Na
ti
o
n
a
l
Un
i
v
e
rsity
o
f
M
a
lay
sia
(UK
M
).
S
in
c
e
c
o
m
p
letin
g
h
is
e
d
u
c
a
ti
o
n
,
h
e
h
a
s
b
e
e
n
wo
r
k
in
g
a
s
a
c
o
m
p
u
t
e
r
sc
ien
c
e
re
se
a
rc
h
e
r
a
n
d
a
fu
ll
-
ti
m
e
a
c
c
o
ss
iate
p
ro
fe
ss
o
r
in
th
e
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter
S
c
ien
c
e
a
t
S
EU
Un
i
v
e
rsity
in
S
a
u
d
i
Ara
b
ia.
His
re
se
a
rc
h
in
tere
sts
in
c
l
u
d
e
a
rti
ficia
l
i
n
telli
g
e
n
c
e
,
m
a
c
h
in
e
lea
rn
i
n
g
a
n
d
o
p
ti
m
iza
ti
o
n
,
c
o
m
p
u
ter
v
isi
o
n
,
d
e
e
p
lea
rn
in
g
,
a
n
d
th
e
In
tern
e
t
o
f
Th
in
g
s.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
t.
a
b
o
a
i
n
@s
e
u
.
e
d
u
.
sa
.
Wa
lee
d
Abd
e
l
K
a
r
i
m
Ab
u
a
i
n
Wale
e
d
A
b
d
e
l
Ka
rim
Ab
u
a
in
re
c
e
iv
e
d
h
is P
h
D
in
a
rti
ficia
l
in
telli
g
e
n
c
e
in
2
0
1
6
fr
o
m
th
e
Artifi
c
ial
In
telli
g
e
n
c
e
T
e
c
h
n
o
l
o
g
y
Ce
n
ter
o
f
t
h
e
Na
ti
o
n
a
l
Un
i
v
e
rsity
o
f
M
a
lay
sia
,
De
p
a
rtme
n
t
o
f
Co
m
p
u
t
e
r
S
c
ien
c
e
(UK
M
).
S
in
c
e
2
0
1
6
,
h
e
is
c
u
rre
n
tl
y
wo
r
k
in
g
a
s
a
fu
ll
-
ti
m
e
a
c
c
o
ss
iate
p
ro
fe
ss
o
r
i
n
t
h
e
De
p
a
r
tme
n
t
o
f
Co
m
p
u
ter
S
c
ien
c
e
a
t
Taib
a
h
Un
i
v
e
rsity
,
A
p
p
li
e
d
C
o
ll
e
g
e
,
S
a
u
d
i
Ara
b
ia.
His
re
se
a
rc
h
i
n
tere
sts
i
n
c
lu
d
e
a
rti
ficia
l
in
telli
g
e
n
c
e
,
m
a
c
h
in
e
lea
rn
in
g
a
n
d
o
p
ti
m
iza
ti
o
n
,
d
a
ta
sc
ien
c
e
,
c
o
m
p
u
ter
v
isi
o
n
,
d
e
e
p
lea
rn
i
n
g
,
a
n
d
t
h
e
in
ter
n
e
t
o
f
th
i
n
g
s.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
wa
b
u
a
in
@t
a
ib
a
h
u
.
e
d
u
.
sa
.
Evaluation Warning : The document was created with Spire.PDF for Python.