I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
37
,
No
.
3
,
Ma
r
ch
20
25
,
p
p
.
1
5
8
8
~1
595
I
SS
N:
2
502
-
4
7
52
,
DOI
:
1
0
.
1
1
5
9
1
/ijee
cs
.v
37.
i
3
.
p
p
1
5
8
8
-
1
5
9
5
1588
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
Co
nv
o
lutiona
l
neura
l net
wo
rk
f
o
r
speech emo
tion
re
co
g
nition in
th
e Mo
ro
ccan
Ar
a
bic
dia
lect
la
ng
u
a
g
e
So
ufiy
a
n O
ua
li
,
Sa
id E
l G
a
ro
ua
ni
D
e
p
a
r
t
me
n
t
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
,
S
i
d
i
M
o
h
a
me
d
B
e
n
A
b
d
e
l
l
a
h
U
n
i
v
e
r
si
t
y
,
F
e
z
,
M
o
r
o
c
c
o
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Mar
28
,
2
0
2
4
R
ev
is
ed
Oct
20
,
2
0
2
4
Acc
ep
ted
Oct
28
,
2
0
2
4
Ex
trac
ti
n
g
th
e
sp
e
a
k
e
r'
s
e
m
o
ti
o
n
a
l
sta
te
h
a
s
b
e
c
o
m
e
a
n
a
c
ti
v
e
re
se
a
rc
h
to
p
ic
late
ly
d
u
e
to
th
e
d
e
m
a
n
d
fo
r
m
o
r
e
h
u
m
a
n
i
n
tera
c
ti
v
e
a
p
p
li
c
a
ti
o
n
s.
Th
is
field
o
f
re
se
a
rc
h
h
a
s
n
o
ted
sig
n
ifi
c
a
n
t
a
d
v
a
n
c
e
m
e
n
t,
e
sp
e
c
ially
i
n
t
h
e
En
g
li
sh
lan
g
u
a
g
e
,
o
win
g
to
t
h
e
a
v
a
il
a
b
il
it
y
o
f
m
a
ss
iv
e
s
p
e
e
c
h
-
lab
e
led
c
o
rp
o
ra
.
Ho
we
v
e
r,
t
h
e
p
ro
g
re
ss
o
f
a
n
a
lo
g
o
u
s
m
e
th
o
d
o
lo
g
ies
i
n
t
h
e
Ara
b
ic
la
n
g
u
a
g
e
is
stil
l
in
it
s
i
n
fa
n
c
y
sta
g
e
s.
Th
is
p
a
p
e
r
p
re
se
n
ts
a
n
e
w
m
a
ss
iv
e
n
a
tu
ra
l
sp
e
e
c
h
e
m
o
ti
o
n
d
a
tas
e
t
a
n
d
a
sp
e
e
c
h
re
c
o
g
n
i
ti
o
n
m
o
d
e
l
f
o
r
th
e
M
o
ro
c
c
a
n
Ara
b
ic
lan
g
u
a
g
e
.
F
o
u
r
p
r
ima
ry
e
m
o
ti
o
n
l
a
b
e
ls
we
re
se
le
c
ted
:
h
a
p
p
y
,
sa
d
,
a
n
g
r
y
,
a
n
d
n
e
u
tral
.
Va
rio
u
s
sp
e
c
tral
fe
a
tu
re
s,
su
c
h
a
s
th
e
mel
-
fre
q
u
e
n
c
y
c
e
p
stra
l
c
o
e
fficie
n
t
(M
F
CC),
we
re
e
x
tra
c
ted
a
n
d
tes
ted
t
o
d
e
term
in
e
t
h
e
o
p
ti
m
a
l
fe
a
tu
re
c
o
m
b
in
a
ti
o
n
.
A
c
o
n
v
o
l
u
t
io
n
a
l
n
e
u
ra
l
n
e
two
r
k
s
(CNN
s)
m
o
d
e
l
wa
s
b
u
il
t
a
n
d
train
e
d
o
n
o
u
r
d
a
tas
e
t.
T
h
e
re
su
lt
s
we
re
c
o
m
p
a
re
d
b
e
twe
e
n
s
p
e
c
tral
fe
a
tu
re
s
in
d
i
v
id
u
a
ll
y
a
n
d
c
o
m
b
i
n
e
d
with
th
e
CNN
m
o
d
e
l
re
su
lt
in
g
i
n
t
h
e
se
lec
ti
o
n
o
f
M
F
CC,
r
o
o
t
-
mean
-
s
q
u
a
re
(RM
S
),
m
el
-
sc
a
led
sp
e
c
tro
g
ra
m
,
a
n
d
sp
e
c
tral,
a
s
o
p
t
ima
l
sp
e
c
tral
fe
a
tu
re
s
fo
r
o
u
r
d
a
tas
e
t.
T
h
is
se
lec
ti
o
n
y
ield
e
d
sig
n
ifi
c
a
n
t
re
su
lt
s,
wit
h
a
n
a
c
c
u
ra
c
y
o
f
9
9
.
5
5
%
f
o
r
e
m
o
t
io
n
re
c
o
g
n
i
ti
o
n
,
o
u
t
p
e
rfo
rm
in
g
t
h
e
e
x
isti
n
g
re
se
a
rc
h
.
K
ey
w
o
r
d
s
:
Ar
ab
ic
SER
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
Featu
r
e
ex
tr
ac
tio
n
Sig
n
al
p
r
o
ce
s
s
in
g
Sp
ee
ch
em
o
tio
n
r
ec
o
g
n
itio
n
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
:
So
u
f
iy
an
Ou
ali
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
,
Facu
lty
o
f
Scien
ce
,
Sid
i
Mo
h
am
ed
B
en
Ab
d
ellah
Un
iv
er
s
ity
Fez,
Mo
r
o
cc
o
E
m
ail: so
u
f
iy
an
e.
o
u
ali@
u
s
m
b
a.
ac
.
m
a
1.
I
NT
RO
D
UCT
I
O
N
Fro
m
th
e
ea
r
ly
d
ec
ad
es
o
f
m
e
ch
an
izatio
n
,
h
u
m
an
s
wer
e
in
ter
ested
in
b
u
ild
in
g
m
ac
h
i
n
es
to
f
ac
ilit
ate
th
eir
liv
es.
T
h
er
ef
o
r
e,
th
e
a
d
v
an
ce
m
en
t
an
d
in
n
o
v
atio
n
o
f
m
an
y
au
to
m
atic
m
ac
h
in
es
wer
e
b
u
ilt
to
r
ep
lace
th
e
d
aily
r
o
u
tin
e
jo
b
s
o
f
h
u
m
an
s
[
1
]
.
Hen
ce
,
th
ey
b
u
i
lt
s
m
ar
t
cities,
h
o
m
es,
ca
r
s
,
an
d
p
h
o
n
es.
No
wad
ay
s
,
r
esear
ch
er
s
ar
e
in
te
r
ested
in
b
u
ild
in
g
m
o
r
e
ad
v
an
ce
d
m
ac
h
in
es
th
at
m
im
ic
h
u
m
an
k
i
n
d
'
s
th
in
k
in
g
an
d
u
n
d
er
s
tan
d
i
n
g
,
wh
ich
r
ev
o
lu
t
io
n
ized
th
e
h
u
m
an
-
c
o
m
p
u
ter
in
ter
ac
tio
n
(
HC
I
)
[
2
]
f
ield
.
HC
I
h
as
em
er
g
ed
to
war
d
s
in
n
o
v
atio
n
,
d
esig
n
,
a
n
d
co
n
s
tr
u
ctio
n
o
f
n
ew
k
in
d
s
o
f
in
f
o
r
m
atio
n
an
d
in
ter
ac
tio
n
tech
n
o
lo
g
y
with
h
u
m
an
-
lik
e
in
tellig
en
ce
[
3
]
b
y
u
s
in
g
ad
v
an
ce
d
tech
n
i
q
u
es
s
u
ch
as
s
p
ee
ch
r
ec
o
g
n
itio
n
(
SR
)
.
SR
i
s
a
s
u
b
-
f
iled
o
f
HC
I
in
ter
ested
in
b
u
ild
in
g
s
m
ar
t
en
titi
es
th
at
ca
n
u
n
d
er
s
tan
d
an
d
in
te
r
p
r
et
v
o
ices
h
u
m
a
n
ly
u
s
in
g
p
h
o
n
etic
an
d
p
ar
alin
g
u
is
tic
v
o
ice
d
ata,
v
er
b
al
an
d
n
o
n
-
v
er
b
al
elem
e
n
ts
s
u
ch
as
v
o
ice
-
to
n
,
p
itch
,
an
d
s
p
ee
d
[
4
]
,
[
5
]
.
R
esear
ch
er
s
in
r
ef
[
6
]
h
ig
h
lig
h
ted
th
at
co
n
v
er
s
atio
n
is
h
i
g
h
l
y
af
f
ec
te
d
b
y
th
e
u
n
d
er
s
tan
d
i
n
g
o
f
p
ar
alin
g
u
is
tic
f
ea
tu
r
es.
T
h
e
r
ef
o
r
e
,
a
n
ew
f
iel
d
o
f
s
tu
d
y
h
as
em
e
r
g
ed
,
s
p
ee
c
h
-
em
o
tio
n
r
ec
o
g
n
itio
n
(
SER)
wh
ich
is
a
s
u
b
-
f
ield
o
f
SR
in
ter
ested
in
e
x
tr
ac
tin
g
th
e
s
p
ea
k
er
'
s
em
o
tio
n
al
s
tate
b
ased
o
n
s
p
ee
ch
s
ig
n
al
p
r
o
p
er
ties
[
7
]
.
On
e
o
f
t
h
e
m
ain
q
u
esti
o
n
s
th
at
d
r
ew
t
h
e
a
tten
tio
n
o
f
r
esear
ch
er
s
in
th
e
S
E
R
f
ield
is
wh
at
ar
e
th
e
im
p
o
r
tan
t
v
o
ice
f
ea
tu
r
es
th
at
id
en
tify
a
s
p
ea
k
er
'
s
em
o
tio
n
al
s
tate.
Mo
r
eo
v
e
r
,
wh
at
m
ak
es
SER
im
p
o
r
tan
t
to
th
e
HC
I
f
ield
is
th
at
em
p
lo
y
in
g
SER
will
en
ab
le
c
o
m
p
u
ter
s
t
o
tak
e
a
s
u
itab
le
a
ctio
n
ac
co
r
d
in
g
to
th
e
s
p
ea
k
e
r
'
s
em
o
tio
n
al
s
tate
wh
ich
is
a
h
u
m
an
-
lik
e
ch
ar
ac
t
er
is
tic.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
C
o
n
vo
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
fo
r
s
p
ee
ch
emo
tio
n
r
ec
o
g
n
itio
n
in
th
e
Mo
r
o
cc
a
n
A
r
a
b
ic
…
(
S
o
u
fiya
n
Ou
a
li
)
1589
T
h
e
p
r
im
a
r
y
f
r
am
ewo
r
k
f
o
r
co
n
s
tr
u
ctin
g
a
SER
s
y
s
tem
is
as
f
o
llo
ws:
(
a)
d
ataset
b
u
ild
in
g
.
SER
d
atasets
ar
e
ca
teg
o
r
ized
i
n
to
t
h
r
ee
k
in
d
s
[
8
]
;
n
atu
r
al
o
r
s
p
o
n
tan
eo
u
s
s
p
ee
ch
d
atasets
,
wh
i
ch
co
n
tain
n
atu
r
a
l
an
d
au
th
en
tic
em
o
tio
n
s
;
a
cted
d
atasets
,
co
m
p
o
s
ed
o
f
r
ec
o
r
d
in
g
s
b
y
a
p
r
o
f
ess
io
n
al
v
o
ice
ac
to
r
;
an
d
e
licited
d
ataset,
wh
ich
in
v
o
lv
es
in
d
u
ci
n
g
s
p
ec
if
ic
e
m
o
tio
n
s
i
n
a
p
er
s
o
n
to
r
ec
o
r
d
t
h
eir
s
p
ee
ch
.
(
b
)
f
ea
tu
r
e
e
x
tr
ac
tio
n
.
Sp
ee
ch
is
a
co
n
tin
u
o
u
s
s
ig
n
al
co
n
tain
in
g
g
lo
b
al
an
d
lo
ca
l
in
f
o
r
m
atio
n
[
9
]
.
T
h
er
ef
o
r
e
,
f
ea
tu
r
e
ex
tr
ac
tio
n
is
th
e
p
h
ase
o
f
ex
t
r
ac
tin
g
n
u
m
e
r
ical
d
ata
f
r
o
m
s
p
ee
ch
u
s
in
g
s
ig
n
a
l
p
r
o
p
er
ties
wh
ich
ar
e
ca
teg
o
r
ized
in
to
4
class
es:
p
r
o
s
o
d
i
c
f
e
a
t
u
r
e
s
,
s
p
e
c
t
r
a
l
f
e
a
t
u
r
e
s
,
v
o
i
c
e
q
u
a
l
i
t
y
f
e
a
t
u
r
e
s
,
a
n
d
t
e
a
g
e
r
e
n
e
r
g
y
o
p
e
r
a
t
o
r
(
T
E
O
)
b
a
s
e
d
f
e
a
t
u
r
e
s
[
8
]
.
(
c)
c
lass
if
icatio
n
a
lg
o
r
ith
m
,
a
f
ter
co
llectin
g
a
n
d
la
b
elin
g
t
h
e
d
ataset
it
is
f
ed
in
t
o
a
cla
s
s
if
ier
to
lear
n
th
e
p
atter
n
s
an
d
co
n
n
ec
tio
n
s
b
etwe
en
ea
ch
u
tter
an
ce
to
p
r
ed
i
ct
n
ew
o
r
u
n
lab
eled
d
ata.
B
o
th
m
ac
h
in
e
lear
n
in
g
(
ML
)
,
e.
g
.
,
R
F,
DT
,
KNN,
an
d
d
ee
p
lear
n
in
g
,
e.
g
.
,
R
NN,
L
STM
,
m
o
d
els
ar
e
u
s
ed
f
o
r
t
h
is
task
,
th
e
ch
o
ice
b
etwe
en
ML
an
d
DL
m
o
d
els d
ep
en
d
s
o
n
v
ar
io
u
s
f
ac
to
r
s
s
u
ch
as th
e
s
ize,
ty
p
e,
an
d
len
g
th
o
f
th
e
d
ataset.
T
h
e
Ar
ab
ic
lan
g
u
ag
e
ca
n
b
e
d
iv
id
ed
in
to
th
r
ee
ca
teg
o
r
ies.
C
lass
ic
Ar
ab
ic,
an
o
f
f
icial
lan
g
u
ag
e
o
f
all
Ar
ab
s
an
d
th
e
lan
g
u
ag
e
o
f
t
h
e
Qu
r
an
.
Mo
d
e
r
n
s
tan
d
ar
d
Ar
a
b
ic
(
MSA)
,
is
a
f
o
r
m
al
co
m
m
u
n
icatio
n
lan
g
u
ag
e
co
m
m
o
n
l
y
u
s
ed
in
n
ews
an
d
ac
ad
em
ic
p
a
p
er
s
.
C
o
llo
q
u
ial
o
r
d
ialec
tal
A
r
ab
ic
lan
g
u
ag
e
is
u
s
ed
in
d
aily
o
r
f
am
iliar
co
n
v
e
r
s
atio
n
s
an
d
it
d
ep
en
d
s
o
n
r
eg
i
o
n
al
v
ar
iatio
n
s
.
Desp
ite
th
e
ad
v
an
ce
m
en
ts
i
n
th
e
f
ield
o
f
S
E
R
,
p
ar
ticu
lar
ly
in
E
n
g
lis
h
,
Ar
ab
i
c
r
esear
ch
is
s
till
in
its
ea
r
ly
s
tag
es.
T
h
is
ca
n
b
e
at
tr
ib
u
ted
to
v
ar
io
u
s
f
ac
to
r
s
,
in
clu
d
in
g
th
e
s
ca
r
city
o
f
well
-
s
tr
u
ctu
r
ed
d
atasets
.
I
n
th
e
s
tate
-
of
-
th
e
-
a
r
t
liter
atu
r
e,
we
f
o
u
n
d
th
at
o
n
l
y
th
r
ee
SER r
esear
ch
es we
r
e
co
n
d
u
ct
ed
o
n
d
ialec
tal
Ar
ab
ic
in
s
ec
tio
n
2
.
R
ec
en
tly
,
m
o
r
e
in
ter
est
h
as
b
ee
n
d
ir
ec
ted
to
war
d
s
t
h
e
Ar
ab
ic
lan
g
u
ag
e
.
Hif
n
y
an
d
Ali
[
1
0
]
au
t
h
o
r
s
p
r
o
p
o
s
ed
a
n
ew
d
ee
p
n
e
u
r
al
n
etwo
r
k
atten
tio
n
-
b
ased
ar
c
h
itectu
r
e
co
n
s
is
tin
g
o
f
C
NN
-
B
L
STM
-
DNN
an
d
tr
ain
ed
it
o
n
th
e
KSUE
m
o
tio
n
s
d
ataset
[
1
1
]
f
r
o
m
wh
ich
th
e
y
ex
tr
ac
ted
1
3
MFC
C
s
p
ec
tr
al
f
ea
tu
r
es
to
b
u
ild
an
Ar
ab
ic
SER
ab
le
t
o
id
e
n
tify
f
iv
e
em
o
tio
n
lab
els
(
i.e
.
,
n
eu
tr
al,
h
a
p
p
y
,
s
ad
,
s
u
r
p
r
is
ed
,
an
d
q
u
esti
o
n
in
g
)
.
E
m
p
lo
y
in
g
th
is
ar
ch
itect
e
n
a
b
led
th
e
r
esear
ch
e
r
s
to
ac
h
ie
v
e
an
ac
cu
r
ac
y
o
f
8
7
.
2
%.
A
h
m
ed
[
1
2
]
b
u
ilt
a
n
Ar
ab
ic
SER
b
ased
o
n
th
e
B
AVE
D
d
ataset
[
1
3
]
to
i
d
en
ti
f
y
th
r
ee
em
o
ti
o
n
s
tates,
lo
w
em
o
tio
n
(
tire
d
o
r
ex
h
au
s
ted
)
,
n
eu
t
r
al
em
o
tio
n
,
an
d
h
ig
h
em
o
tio
n
(
p
o
s
itiv
e
o
r
n
eg
ati
v
e
em
o
tio
n
s
,
h
ap
p
i
n
e
s
s
,
jo
y
,
s
ad
n
ess
,
o
r
an
g
er
)
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
t
ested
b
o
th
W
av
2
v
ec
2
.
0
an
d
H
u
B
E
R
T
f
o
r
th
e
f
ea
t
u
r
e
ex
tr
ac
t
io
n
p
h
ase
a
n
d
th
e
n
th
e
o
u
tp
u
t
was
f
ed
to
ML
P
c
lass
if
ier
s
to
p
r
ed
ict
th
e
em
o
tio
n
lab
el.
T
h
e
r
esu
lt
s
h
o
wed
th
at
W
av
2
v
ec
2
.
0
ac
h
iev
ed
8
9
%
ac
cu
r
ac
y
,
s
u
r
p
ass
in
g
Hu
B
E
R
T
,
wh
ich
ac
h
ie
v
ed
an
ac
cu
r
ac
y
o
f
8
4
%.
I
n
a
s
im
ilar
s
tu
d
y
b
y
K
lay
lat
et
a
l.
[
1
4
]
r
esear
ch
er
s
ex
tr
ac
ted
o
n
ly
th
e
p
r
o
s
o
d
ic
f
e
atu
r
e
f
r
o
m
an
Ar
ab
ic
d
ataset
th
ey
b
u
ilt
m
an
u
ally
wh
ich
co
n
s
is
ts
o
f
3
em
o
tio
n
class
es
(
i.e
.
,
h
ap
p
y
,
an
g
r
y
,
s
u
r
p
r
is
ed
)
,
an
d
test
ed
it
with
a
t
h
ir
ty
-
f
iv
e
class
if
icatio
n
s
m
o
d
el
to
f
in
d
o
u
t
th
at
th
e
b
est
r
esu
lt
o
b
tain
ed
u
s
in
g
th
e
s
eq
u
en
tial
m
in
im
al
o
p
tim
izatio
n
with
an
ac
cu
r
ac
y
o
f
9
5
.
5
2
%.
I
n
Ar
ab
ic
d
ialec
ts
,
th
r
ee
r
esear
ch
es
wer
e
co
n
d
u
cted
I
n
[
1
2
]
r
esear
ch
er
s
b
u
ilt
an
SER
s
y
s
tem
f
o
r
th
e
Sau
d
i
Ar
a
b
ic
d
ialec
t
to
id
en
tif
y
f
o
u
r
em
o
ti
o
n
lab
els
(
i.e
.
,
h
a
p
p
y
,
s
ad
,
a
n
g
r
y
,
n
eu
t
r
al)
,
th
e
y
cr
ea
ted
a
d
ataset
m
an
u
ally
f
r
o
m
wh
ich
th
ey
e
x
tr
ac
ted
an
d
test
ed
th
r
ee
s
p
ec
tr
al
f
ea
tu
r
es
(
i.e
.
,
MFC
C
,
m
el
s
p
ec
tr
o
g
r
am
,
an
d
s
p
ec
tr
al
co
n
tr
ast)
in
d
iv
id
u
ally
an
d
co
m
b
i
n
ed
,
with
th
r
ee
class
if
icatio
n
m
o
d
els
(
i.e
.
,
SVM,
KNN,
ML
P),
th
e
r
esu
lt
s
h
o
wed
th
at
th
e
KNN
p
er
f
o
r
m
ed
b
etter
with
th
e
co
m
b
in
atio
n
o
f
MF
CC
an
d
m
el
s
p
ec
tr
o
g
r
am
with
an
ac
cu
r
ac
y
o
f
6
8
.
5
7
%,
a
n
d
t
h
e
SVM
an
d
ML
P
m
o
d
el
p
er
f
o
r
m
e
d
b
etter
with
ac
cu
r
ac
ies
o
f
7
7
.
1
4
%
an
d
7
1
.
4
3
%,
r
esp
ec
t
iv
ely
.
Ab
d
el
-
Ham
id
[
1
5
]
r
es
ea
r
ch
er
s
b
u
ilt
a
s
em
i
-
n
eu
tr
al
E
g
y
p
tian
Ar
a
b
ic
d
ataset
co
n
s
is
tin
g
o
f
f
o
u
r
em
o
tio
n
s
(
i.e
.
,
h
a
p
p
y
,
s
ad
,
a
n
g
r
y
,
n
eu
tr
al)
ca
lled
‘
E
YASE
’
.
Pro
s
o
d
ic
,
s
p
ec
tr
al,
an
d
wav
elet
f
ea
tu
r
es
wer
e
ex
tr
ac
ted
an
d
co
m
p
ar
ed
with
d
if
f
er
en
t
co
m
b
in
atio
n
s
with
SVM
an
d
KNN
class
if
icatio
n
m
o
d
els.
T
h
e
ex
p
er
im
en
tal
r
esu
lts
s
h
o
wed
th
at
SVM
o
u
tp
er
f
o
r
m
e
d
KNN,
ac
h
iev
in
g
an
ac
cu
r
ac
y
o
f
9
5
%
f
o
r
d
etec
tin
g
em
o
tio
n
f
r
o
m
m
ale
v
o
ices,
8
4
.
2
% f
o
r
f
em
ale
v
o
ices,
a
n
d
9
0
.
7
%
f
o
r
b
o
th
g
e
n
d
er
s
.
An
g
er
em
o
tio
n
was
f
o
u
n
d
to
b
e
th
e
ea
s
iest
class
to
d
etec
t.
Mu
s
taf
a
et
a
l.
[
1
6
]
r
esear
ch
was
co
n
d
u
cte
d
o
n
th
e
Alg
er
ian
Ar
ab
ic
d
ialec
t.
T
h
e
r
esear
ch
er
s
u
s
ed
th
e
MFC
C
s
p
ec
tr
al
f
ea
tu
r
e
an
d
test
ed
it
with
twelv
e
class
if
icatio
n
m
o
d
els
o
n
th
eir
m
an
u
ally
b
u
ilt
d
ataset
to
d
et
ec
t
f
o
u
r
em
o
tio
n
lab
els
(
i.e
.
,
h
ap
p
y
,
s
ad
,
an
g
r
y
,
n
eu
tr
al)
,
th
e
r
es
u
lt
s
h
o
wed
t
h
at
th
e
b
est
ac
cu
r
ac
y
o
b
tain
e
d
with
L
STM
-
C
NN
m
o
d
el
w
ith
an
ac
cu
r
ac
y
o
f
9
3
.
3
4
%.
Desp
ite
th
e
n
u
m
er
o
u
s
ad
v
a
n
ce
m
en
ts
in
SER
s
y
s
tem
s
f
o
r
th
e
Ar
ab
ic
lan
g
u
ag
e
,
a
s
ig
n
if
ican
t
ch
allen
g
e
r
em
ai
n
s
u
n
ad
d
r
ess
ed
.
T
h
e
v
ast
d
iv
er
s
ity
with
i
n
Ar
ab
ic
d
ialec
ts
,
wh
i
ch
v
a
r
ies
g
r
ea
tly
d
u
e
to
r
eg
io
n
al
ch
a
r
ac
ter
is
tics
,
p
o
s
es
a
s
u
b
s
tan
tial
o
b
s
tacle
.
T
h
ese
v
ar
iatio
n
s
h
in
d
e
r
th
e
g
e
n
er
al
izatio
n
o
f
ex
is
tin
g
SER
m
o
d
els,
r
en
d
er
in
g
th
em
in
ef
f
ec
tiv
e
ac
r
o
s
s
d
if
f
er
en
t
d
i
alec
ts
.
C
o
n
s
eq
u
en
tly
,
th
e
cr
ea
tio
n
o
f
s
p
ec
ialize
d
SER
s
y
s
tem
s
f
o
r
ea
ch
d
ialec
ta
l
Ar
ab
ic
is
n
o
t
ju
s
t
im
p
o
r
tan
t
b
u
t
it
is
im
p
er
ativ
e.
T
h
e
r
ef
o
r
e
,
a
t
h
o
r
o
u
g
h
r
e
v
iew
o
f
cu
r
r
en
t
r
esear
ch
r
ev
ea
ls
a
g
lar
in
g
g
ap
:
n
o
r
esear
ch
er
h
as
y
et
tack
led
th
e
u
n
iq
u
e
c
o
m
p
lex
ities
o
f
th
e
Mo
r
o
cc
an
Ar
ab
ic
d
ialec
t.
T
h
i
s
o
v
er
s
ig
h
t
leav
es
a
cr
it
ical
v
o
id
in
t
h
e
f
ield
an
d
h
ig
h
lig
h
ts
an
u
r
g
en
t
n
ee
d
f
o
r
f
o
cu
s
ed
r
esear
ch
.
T
h
e
m
ain
c
o
n
tr
ib
u
tio
n
o
f
o
u
r
p
a
p
er
i
n
th
e
SER
f
ield
is
s
u
m
m
a
r
ized
as
f
o
llo
ws:
i)
p
r
esen
t
th
e
f
ir
s
t
s
p
ee
ch
em
o
tio
n
d
ata
s
et
f
o
r
th
e
Mo
r
o
cc
an
Ar
a
b
ic
lan
g
u
ag
e
(
MA
DE
S)
,
ii)
b
u
ild
in
g
an
S
E
R
s
y
s
tem
ab
le
to
ex
tr
ac
t
th
e
em
o
tio
n
al
s
tate
f
r
o
m
th
e
s
p
ee
ch
o
f
th
e
Mo
r
o
cc
an
s
p
ea
k
er
s
,
an
d
ii
i)
c
o
n
s
tr
u
ctin
g
an
ef
f
icien
t
d
ee
p
-
lear
n
in
g
m
o
d
el
th
at
o
u
tp
er
f
o
r
m
s
th
e
ex
is
tin
g
r
esear
ch
in
th
e
SER
f
ield
.
T
h
e
r
est
o
f
th
e
p
a
p
er
is
o
r
g
an
ized
as
f
o
llo
ws:
I
n
s
ec
tio
n
2
,
we
p
r
esen
t
th
e
Me
th
o
d
o
lo
g
y
o
f
o
u
r
wo
r
k
.
T
h
e
ex
p
e
r
im
e
n
t,
d
is
cu
s
s
io
n
,
an
d
r
esu
lts
ar
e
p
r
esen
ted
in
s
ec
tio
n
3
.
Fin
ally
,
s
ec
tio
n
4
co
n
clu
d
e
s
th
e
p
ap
er
a
n
d
d
is
cu
s
s
es f
u
tu
r
e
wo
r
k
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
3
,
Ma
r
ch
20
25
:
1
588
-
1
5
9
5
1590
2.
M
E
T
H
O
D
2
.
1
.
Da
t
a
s
et
bu
ild
ing
I
n
th
is
wo
r
k
,
we
c
r
ea
ted
a
n
a
tu
r
al
em
o
tio
n
s
p
ee
c
h
d
ataset
f
o
r
th
e
MA
DE
S
.
T
o
k
ee
p
th
e
em
o
tio
n
al
n
atu
r
aln
ess
o
f
o
u
r
c
o
r
p
u
s
,
th
e
d
ata
wer
e
co
llected
f
r
o
m
p
o
p
u
lar
Mo
r
o
cc
an
r
ea
lity
T
V
s
h
o
w
s
,
r
ad
io
p
r
o
g
r
am
s
,
p
o
d
ca
s
ts
,
an
d
in
ter
v
iews.
A
g
r
o
u
p
o
f
au
d
io
a
n
d
v
id
e
o
s
wer
e
co
n
d
u
cted
to
s
elec
t
o
n
ly
m
ea
n
in
g
f
u
l
o
n
es
th
at
in
clu
d
e
clea
r
em
o
tio
n
al
r
ef
er
e
n
ce
s
.
Su
b
s
eq
u
en
tly
,
all
th
e
d
a
ta
ar
e
lab
eled
an
d
co
n
v
er
ted
i
n
to
au
d
io
u
s
in
g
th
e
W
AV
f
o
r
m
at.
T
h
e
f
in
al
d
ata
s
et
co
n
s
i
s
ts
o
f
1
,
5
0
5
r
ec
o
r
d
s
v
ar
ied
f
r
o
m
th
r
ee
t
o
ten
s
ec
o
n
d
s
,
b
ased
o
n
th
e
s
en
ten
ce
len
g
th
.
B
ased
o
n
r
esear
ch
in
[
1
6
]
C
o
m
m
o
n
ly
,
ea
c
h
s
p
ee
ch
ca
n
e
x
p
r
ess
o
n
e
o
f
th
e
f
o
u
r
em
o
tio
n
s
:
h
ap
p
in
ess
,
an
g
e
r
,
s
ad
n
ess
,
o
r
n
eu
tr
ality
.
T
h
e
r
ef
o
r
e,
we
lab
el
ed
ea
ch
r
ec
o
r
d
with
o
n
e
o
f
t
h
ese
em
o
tio
n
lab
els.
T
h
e
r
ec
o
r
d
lab
elin
g
was
co
n
d
u
cted
s
ep
ar
ately
b
y
two
n
ativ
e
Mo
r
o
cc
an
s
p
ea
k
e
r
s
.
On
ly
r
e
co
r
d
s
th
at
r
ec
eiv
ed
th
e
s
am
e
lab
el
f
r
o
m
b
o
th
ass
ess
o
r
s
,
in
d
icatin
g
a
clea
r
r
e
p
r
esen
tatio
n
o
f
th
e
em
o
tio
n
,
wer
e
in
clu
d
e
d
in
th
e
d
atab
ase.
T
o
en
s
u
r
e
th
e
r
ep
r
esen
tativ
en
ess
o
f
th
e
d
ataset
we
in
clu
d
ed
au
d
io
s
am
p
les
f
r
o
m
b
o
th
g
en
d
er
s
ac
r
o
s
s
v
ar
io
u
s
ag
e
g
r
o
u
p
s
,
co
m
p
r
is
in
g
7
8
5
r
ec
o
r
d
s
o
f
m
ales
an
d
7
2
0
r
ec
o
r
d
s
o
f
f
em
ales.
Mo
r
e
o
v
er
,
to
r
e
d
u
ce
th
e
r
is
k
o
f
o
v
er
f
itti
n
g
th
e
s
am
p
l
e
s
ize
f
o
r
ea
ch
ca
teg
o
r
y
is
ap
p
r
o
x
im
ately
eq
u
al.
Fu
r
th
er
m
o
r
e,
ea
c
h
r
ec
o
r
d
in
clu
d
es
th
r
ee
p
iece
s
o
f
in
f
o
r
m
atio
n
:
th
e
em
o
tio
n
al
s
tate,
a
g
e,
an
d
g
en
d
e
r
o
f
th
e
s
p
ea
k
er
.
T
h
u
s
,
it
ca
n
also
b
e
u
tili
ze
d
f
o
r
g
en
d
er
s
p
ee
ch
r
ec
o
g
n
itio
n
o
r
ag
e
id
en
tific
atio
n
.
T
o
o
u
r
k
n
o
wled
g
e,
th
is
is
th
e
f
i
r
s
t
Data
s
et
f
o
r
em
o
tio
n
r
ec
o
g
n
itio
n
in
t
h
e
M
o
r
o
cc
an
Ar
ab
ic
d
ialec
t.
T
h
e
MA
DE
S
d
ataset
is
av
ailab
le
f
o
r
r
esear
c
h
p
u
r
p
o
s
es
u
p
o
n
r
eq
u
est in
[1
7
]
.
T
ab
le
1
s
u
m
m
ar
izes
th
e
d
ataset
d
is
tr
ib
u
tio
n
.
T
ab
le
1
.
MA
DE
S d
ataset
d
is
tr
ib
u
tio
n
Emo
t
i
o
n
/
g
e
n
d
e
r
M
a
l
e
F
e
mal
e
To
t
a
l
A
g
e
:
[
1
0
-
2
0
]
A
g
e
:
[
2
0
-
4
0
]
A
g
e
:
[
4
0
-
6
0
]
A
g
e
:
[
1
0
-
2
0
]
A
g
e
:
[
2
0
-
4
0
]
A
g
e
:
[
4
0
-
6
0
]
H
a
p
p
y
12
1
4
6
19
21
1
4
2
39
3
7
9
S
a
d
99
62
11
15
1
7
9
15
3
8
1
A
n
g
r
y
63
1
2
5
90
42
22
65
4
0
7
N
e
u
t
r
a
l
68
34
56
56
1
0
2
22
3
3
8
To
t
a
l
2
4
2
3
6
7
1
7
6
1
3
4
4
4
5
1
4
1
1
5
0
5
2
.
2
.
F
e
a
t
ure
ex
t
r
a
ct
io
n
B
u
ild
in
g
an
ef
f
icien
t
m
o
d
el
i
s
h
ig
h
ly
d
ep
e
n
d
en
t
o
n
th
e
q
u
ality
o
f
th
e
t
r
ain
in
g
d
ataset.
Ho
wev
er
,
th
er
e
is
n
o
s
tan
d
ar
d
f
ea
tu
r
e
e
x
tr
ac
tio
n
m
eth
o
d
in
t
h
e
s
tate
-
of
-
th
e
-
a
r
t.
I
n
t
h
is
wo
r
k
,
we
c
h
o
o
s
e
to
ex
t
r
ac
t
th
e
s
p
ec
tr
al
f
ea
tu
r
es
u
s
in
g
t
h
e
L
i
b
r
o
s
a
lib
r
ar
y
[1
8
]
,
as
th
ey
h
a
v
e
p
r
o
v
en
p
r
o
m
is
in
g
r
esu
lts
[
1
2
]
,
[
1
5
]
,
[
1
6
]
.
T
o
ch
o
o
s
e
o
n
ly
th
e
m
o
s
t
s
ig
n
if
ican
t
s
p
ec
tr
al
f
ea
tu
r
es
th
at
ar
e
s
u
itab
le
f
o
r
o
u
r
d
ataset,
we
h
av
e
ex
tr
ac
ted
an
d
test
ed
f
iv
e
s
p
ec
tr
al
f
ea
tu
r
es with
o
u
r
class
if
ier
:
−
Mel
-
f
r
e
q
u
e
n
cy
c
e
p
s
tr
al
c
o
e
f
f
i
cie
n
ts
(
MFC
C
)
,
c
o
n
s
t
it
u
te
a
s
et
o
f
c
o
e
f
f
ici
en
ts
ca
p
t
u
r
i
n
g
t
h
e
s
h
a
p
e
o
f
t
h
e
p
o
we
r
s
p
ec
t
r
u
m
o
f
a
s
o
u
n
d
s
i
g
n
al
[
1
9
]
.
MFC
C
is
w
id
el
y
u
t
iliz
ed
i
n
v
ar
io
u
s
a
p
p
li
ca
t
io
n
s
,
p
a
r
ti
c
u
la
r
l
y
in
v
o
ic
e
s
i
g
n
a
l
p
r
o
ce
s
s
i
n
g
,
s
u
c
h
a
s
s
p
ea
k
e
r
r
ec
o
g
n
iti
o
n
,
v
o
ice
r
e
co
g
n
iti
o
n
,
an
d
g
e
n
d
er
i
d
e
n
t
if
ic
ati
o
n
[
20
].
−
R
o
o
t
-
m
e
an
-
s
q
u
a
r
e
(
R
MS
)
w
h
i
ch
c
o
m
p
u
tes
t
h
e
v
a
lu
e
R
MS
f
o
r
e
ac
h
f
r
am
e,
ei
th
er
f
r
o
m
t
h
e
au
d
i
o
s
am
p
l
e
s
y
o
r
f
r
o
m
a
s
p
ec
t
r
o
g
r
am
.
−
Me
l
s
p
e
ct
r
o
g
r
a
m
is
u
t
ili
ze
d
t
o
co
m
p
u
te
m
el
-
s
c
al
ed
s
p
ec
t
r
o
g
r
am
s
,
a
n
d
f
o
cu
s
in
g
o
n
t
h
e
l
o
w
-
f
r
e
q
u
en
c
y
p
a
r
t
o
f
s
p
e
ec
h
.
−
Sp
e
ct
r
al
f
ea
t
u
r
e
v
ar
ia
n
t
:
s
p
ec
t
r
al
c
en
tr
o
i
d
,
s
p
ec
tr
al
b
a
n
d
wi
d
th
,
s
p
ec
tr
al
c
o
n
tr
ast
,
s
p
ec
tr
al
f
la
tn
ess
,
a
n
d
s
p
e
ct
r
al
r
o
l
l
-
o
f
f
a
r
e
all
e
x
t
r
a
cte
d
.
−
Z
e
r
o
c
r
o
s
s
i
n
g
r
at
e
(
Z
C
R
)
o
f
a
n
a
u
d
i
o
t
im
e
s
er
ies
.
2
.
3
.
Cla
s
s
if
ica
t
io
n
m
o
del
I
n
th
is
wo
r
k
,
a
t
h
o
r
o
u
g
h
in
v
esti
g
atio
n
h
as
b
ee
n
co
n
d
u
cte
d
to
s
elec
t
th
e
m
o
s
t
p
er
tin
e
n
t
class
if
ier
m
o
d
el
f
o
r
o
u
r
d
ataset.
T
h
er
eb
y
,
we
ch
o
o
s
e
to
b
u
ild
an
d
tr
a
in
a
co
n
v
en
tio
n
al
n
e
u
r
al
n
etwo
r
k
(
C
NN)
m
o
d
el
wh
ich
is
a
d
ee
p
-
lear
n
in
g
n
eu
r
al
n
e
two
r
k
k
n
o
wn
f
o
r
its
ab
ilit
y
to
lear
n
co
m
p
lex
p
atter
n
s
an
d
can
r
ec
o
g
n
ize
lo
ca
l a
n
d
g
lo
b
al
ch
ar
ac
te
r
is
tics
o
f
in
p
u
t d
ata,
m
ak
in
g
it su
it
ab
le
f
o
r
SR
ap
p
licatio
n
s
[
2
1
]
-
[2
3
]
.
Mo
r
e
o
v
er
,
th
e
y
h
av
e
s
h
o
wn
g
o
o
d
r
esu
lts
in
th
e
liter
atu
r
e.
Fo
r
ex
am
p
le,
tr
ai
n
in
g
C
NN
m
o
d
e
f
o
r
th
e
SER
s
y
s
tem
en
ab
led
th
e
au
th
o
r
s
in
[2
4
]
to
ac
h
iev
e
a
n
ac
cu
r
ac
y
o
f
8
6
.
0
6
%,
i
n
[2
5
]
r
esear
ch
er
s
ac
h
iev
ed
an
ac
cu
r
a
cy
o
f
7
9
%,
an
d
i
n
[2
6
]
a
u
th
o
r
s
ac
h
ie
v
ed
a
n
F1
-
s
co
r
e
o
f
8
6
.
6
5
%.
T
h
e
C
NN
ar
ch
itectu
r
e
m
o
d
el
d
ev
elo
p
e
d
in
th
is
wo
r
k
co
m
p
r
is
es
(
n
=1
9
)
n
in
etee
n
lay
er
s
,
w
ith
(
n
=1
5
)
f
if
teen
o
f
th
em
b
ein
g
C
NN
lay
er
s
,
as o
u
tlin
ed
in
T
ab
le
2
.
T
o
s
tab
ilize
an
d
ac
ce
ler
ate
th
e
tr
a
in
in
g
o
f
th
e
n
eu
r
al
n
etwo
r
k
,
a
b
atch
n
o
r
m
aliza
tio
n
lay
er
was
ad
d
ed
af
ter
ea
ch
1
D
co
n
v
o
lu
tio
n
al
lay
er
.
T
o
e
n
h
an
ce
th
e
m
o
d
el'
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
C
o
n
vo
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
fo
r
s
p
ee
ch
emo
tio
n
r
ec
o
g
n
itio
n
in
th
e
Mo
r
o
cc
a
n
A
r
a
b
ic
…
(
S
o
u
fiya
n
Ou
a
li
)
1591
g
en
er
aliza
tio
n
ca
p
ab
ilit
y
,
a
d
r
o
p
o
u
t
r
ate
o
f
0
.
2
was
ap
p
lied
b
etwe
en
th
e
1
D
co
n
v
o
lu
tio
n
al
lay
er
,
p
r
o
m
o
tin
g
r
esil
ien
ce
an
d
p
r
ev
e
n
tin
g
o
v
e
r
f
itti
n
g
.
Mo
r
eo
v
er
,
we
ad
d
e
d
a
m
ax
-
p
o
o
lin
g
lay
er
b
etwe
en
th
e
1
D
c
o
n
v
o
lu
tio
n
al
lay
er
to
r
e
d
u
ce
th
e
s
p
atial
d
im
en
s
io
n
s
o
f
th
e
in
p
u
t d
ata.
T
ab
le
2
.
T
h
e
C
NN
ar
ch
itectu
r
e
p
r
o
p
o
s
ed
L
a
y
e
r
s
T
y
p
e
D
e
t
a
i
l
s
1
C
o
n
v
5
1
2
f
i
l
t
e
r
+
k
e
r
n
e
l
s
i
z
e
=
5
+
s
t
r
i
d
e
s
=
1
+
p
a
d
d
i
n
g
=
'
s
a
m
e
'
+
a
c
t
i
v
a
t
i
o
n
=
r
e
l
u
2
B
a
t
c
h
N
o
r
m
a
l
i
z
a
t
i
o
n
d
e
f
a
u
l
t
3
M
a
x
P
o
o
l
1
D
P
o
o
l
s
i
z
e
=
5
+
s
t
r
i
d
e
s
=
2
+
p
a
d
d
i
n
g
=
s
a
m
e
4
C
o
n
v
5
1
2
f
i
l
t
e
r
+
k
e
r
n
e
l
s
i
z
e
=
5
+
s
t
r
i
d
e
s
=
1
+
p
a
d
d
i
n
g
=
'
s
a
m
e
'
+
a
c
t
i
v
a
t
i
o
n
=
r
e
l
u
5
B
a
t
c
h
N
o
r
m
a
l
i
z
a
t
i
o
n
d
e
f
a
u
l
t
6
M
a
x
P
o
o
l
1
D
P
o
o
l
s
i
z
e
=
5
+
s
t
r
i
d
e
s
=
2
+
p
a
d
d
i
n
g
=
s
a
m
e
7
C
o
n
v
2
5
6
f
i
l
t
e
r
+
k
e
r
n
e
l
s
i
z
e
=
5
+
s
t
r
i
d
e
s
=
1
+
p
a
d
d
i
n
g
=
s
a
m
e
+
a
c
t
i
v
a
t
i
o
n
=
r
e
l
u
8
B
a
t
c
h
N
o
r
m
a
l
i
z
a
t
i
o
n
d
e
f
a
u
l
t
9
M
a
x
P
o
o
l
1
D
P
o
o
l
s
i
z
e
=
5
+
s
t
r
i
d
e
s
=
2
+
p
a
d
d
i
n
g
=
s
a
m
e
10
C
o
n
v
2
5
6
f
i
l
t
e
r
+
k
e
r
n
e
l
s
i
z
e
=
3
+
s
t
r
i
d
e
s
=
1
+
p
a
d
d
i
n
g
=
s
a
m
e
+
a
c
t
i
v
a
t
i
o
n
=
r
e
l
u
11
B
a
t
c
h
N
o
r
m
a
l
i
z
a
t
i
o
n
d
e
f
a
u
l
t
12
M
a
x
P
o
o
l
1
D
P
o
o
l
s
i
z
e
=
5
+
s
t
r
i
d
e
s
=
2
+
p
a
d
d
i
n
g
=
s
a
m
e
13
C
o
n
v
1
2
8
f
i
l
t
e
r
+
k
e
r
n
e
l
s
i
z
e
=
3
+
s
t
r
i
d
e
s
=
1
+
p
a
d
d
i
n
g
=
s
a
m
e
+
a
c
t
i
v
a
t
i
o
n
=
r
e
l
u
14
B
a
t
c
h
N
o
r
m
a
l
i
z
a
t
i
o
n
d
e
f
a
u
l
t
15
M
a
x
P
o
o
l
1
D
P
o
o
l
s
i
z
e
=
3
+
s
t
r
i
d
e
s
=
2
+
p
a
d
d
i
n
g
=
s
a
m
e
16
F
l
a
t
t
e
n
d
e
f
a
u
l
t
17
D
e
n
se
5
1
2
n
e
u
r
o
n
s
+
r
e
l
u
a
c
t
i
v
a
t
i
o
n
18
B
a
t
c
h
N
o
r
m
a
l
i
z
a
t
i
o
n
d
e
f
a
u
l
t
19
D
e
n
se
4
n
e
u
r
o
n
s
+
S
o
f
t
M
a
x
a
c
t
i
v
a
t
i
o
n
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
Da
t
a
s
et
prepro
ce
s
s
ing
Af
ter
b
u
ild
i
n
g
t
h
e
MA
DE
S
d
ataset
th
e
t
o
tal
n
u
m
b
er
o
f
r
ec
o
r
d
s
is
1
,
5
0
5
.
R
ec
o
g
n
izin
g
th
e
s
ig
n
if
ican
ce
o
f
ex
ten
s
iv
e
d
ata
in
tr
ain
in
g
an
ef
f
icien
t
m
o
d
e
l,
o
n
e
k
ey
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
e
em
p
lo
y
e
d
is
d
ata
au
g
m
e
n
tatio
n
[
2
7
]
.
T
h
is
p
r
o
ce
s
s
in
v
o
lv
es
cr
ea
tin
g
n
e
w
s
y
n
th
etic
d
ata
s
am
p
les
b
y
in
tr
o
d
u
ci
n
g
s
m
all
p
er
tu
r
b
atio
n
s
to
th
e
in
itial
tr
ai
n
in
g
s
et
th
r
o
u
g
h
th
e
in
jectio
n
o
f
v
ar
i
o
u
s
ef
f
ec
ts
.
T
h
e
ef
f
ec
ts
u
s
ed
in
o
u
r
d
ataset
in
clu
d
e
:
n
o
is
e
in
jectio
n
b
e
ca
u
s
e
in
r
ea
lis
tic
s
ce
n
ar
io
s
s
o
u
n
d
au
d
i
o
s
ig
n
als
f
r
eq
u
en
tly
ex
p
e
r
ien
ce
en
v
ir
o
n
m
en
tal
n
o
is
e,
d
is
to
r
tio
n
s
,
o
r
i
n
ter
f
er
e
n
ce
.
T
h
e
m
o
d
e
l
d
ev
elo
p
s
th
e
ab
ilit
y
to
n
a
v
ig
ate
s
u
ch
s
ce
n
ar
io
s
th
r
o
u
g
h
tr
ain
i
n
g
o
n
d
ata
co
n
tain
in
g
n
o
is
e,
lead
in
g
t
o
m
o
r
e
ac
cu
r
ate
p
r
ed
ictio
n
s
in
r
ea
l
-
wo
r
ld
co
n
d
itio
n
s
.
Sp
ee
d
ch
a
n
g
e
:
I
n
p
r
ac
tical
e
n
v
ir
o
n
m
en
ts
,
s
p
ea
k
i
n
g
s
p
ee
d
s
v
ar
y
.
T
h
e
r
ef
o
r
e
,
two
v
e
r
s
io
n
s
o
f
th
e
o
r
ig
in
al
r
ec
o
r
d
in
g
wer
e
cr
ea
ted
;
o
n
e
w
ith
s
p
ee
d
m
u
ltip
lied
b
y
1
.
2
5
an
d
an
o
th
e
r
with
s
p
ee
d
m
u
ltip
lied
b
y
0
.
8
5
.
T
h
ese
v
alu
es
wer
e
c
h
o
s
en
ca
r
ef
u
lly
to
a
u
g
m
en
t
th
e
d
ata
wh
ile
p
r
eser
v
in
g
th
e
o
r
ig
in
al
s
en
s
e
o
f
th
e
r
ec
o
r
d
i
n
g
.
Sh
if
tin
g
tim
e
:
A
p
r
o
ce
s
s
th
at
en
h
an
ce
s
th
e
d
iv
er
s
ity
o
f
tem
p
o
r
al
asp
ec
ts
in
th
e
tr
ain
in
g
d
ata,
th
er
e
b
y
p
r
o
m
o
tin
g
g
r
ea
ter
r
o
b
u
s
tn
ess
an
d
ad
a
p
tab
ilit
y
in
t
h
e
m
o
d
el.
Pit
ch
ch
an
g
e
:
g
en
er
ate
r
ec
o
r
d
s
b
y
c
h
an
g
in
g
th
e
p
itch
o
f
t
h
e
au
d
i
o
s
ig
n
al.
T
o
m
ain
tain
th
e
s
en
s
e
o
f
t
h
e
o
r
i
g
in
al
r
ec
o
r
d
i
n
g
,
th
e
p
itch
is
ad
j
u
s
ted
b
y
a
f
ac
to
r
o
f
0
.
6
.
T
h
e
ap
p
licatio
n
o
f
d
ata
a
u
g
m
en
tatio
n
,
in
wh
ich
we
im
p
le
m
en
ted
f
iv
e
e
f
f
ec
ts
,
h
as
g
en
e
r
ated
7
,
6
9
9
r
ec
o
r
d
s
,
co
n
tr
ib
u
tin
g
to
t
h
e
m
o
d
el
ac
h
iev
i
n
g
n
o
tab
le
r
esu
lts
,
as
illu
s
tr
ated
in
T
ab
le
3
.
B
ey
o
n
d
cr
ea
ti
n
g
a
s
u
f
f
icien
t
d
ataset,
d
ata
au
g
m
e
n
tatio
n
p
lay
s
a
s
ig
n
if
ican
t
r
o
l
e
in
r
ed
u
cin
g
tr
ain
i
n
g
o
v
e
r
f
it.
As
r
ep
r
esen
ted
in
T
ab
le
3
,
th
e
d
if
f
er
en
c
e
b
etwe
e
n
tr
ain
in
g
a
n
d
v
alid
atio
n
ac
cu
r
ac
y
wh
en
tr
ain
in
g
th
e
m
o
d
el
o
n
th
e
o
r
ig
i
n
al
d
ata
is
6
.
3
%,
an
d
th
e
v
alid
atio
n
lo
s
s
is
5
2
.
5
4
%,
s
ig
n
if
y
in
g
a
s
u
b
s
tan
tial
d
eg
r
ee
o
f
o
v
er
f
itti
n
g
.
I
n
co
n
tr
ast,
wh
en
tr
ain
in
g
th
e
m
o
d
el
with
au
g
m
en
ted
d
ata,
th
e
d
if
f
er
en
ce
b
et
wee
n
tr
ain
in
g
an
d
v
alid
atio
n
ac
cu
r
ac
y
is
0
.
1
1
%,
an
d
th
e
v
alid
atio
n
lo
s
s
is
2
.
5
1
% h
ig
h
lig
h
tin
g
t
h
e
ab
s
en
ce
o
f
o
v
er
f
itti
n
g
a
n
d
g
o
o
d
tr
ain
in
g
.
T
ab
le
3
.
T
h
e
im
p
ac
t o
f
d
ata
a
u
g
m
en
tatio
n
o
n
m
o
d
el
p
er
f
o
r
m
an
ce
Le
a
r
n
i
n
g
r
a
t
e
/
d
a
t
a
s
e
t
W
i
t
h
o
u
t
a
u
g
m
e
n
t
e
d
d
a
t
a
%
W
i
t
h
a
u
g
me
n
t
e
d
d
a
t
a
%
T
r
a
i
n
i
n
g
a
c
c
u
r
a
c
y
8
7
.
8
1
9
9
.
3
3
V
a
l
i
d
a
t
i
o
n
a
c
c
u
r
a
c
y
8
1
.
5
1
9
9
.
2
2
T
r
a
i
n
i
n
g
l
o
ss
3
3
.
0
0
0
2
.
0
4
V
a
l
i
d
a
t
i
o
n
l
o
s
s
5
2
.
5
4
0
2
.
5
1
An
o
th
er
d
ata
p
r
ep
r
o
ce
s
s
in
g
s
tep
in
v
o
l
v
ed
s
tan
d
ar
d
izin
g
th
e
d
ataset,
a
cr
u
cial
p
r
o
ce
d
u
r
e
in
d
ata
an
aly
s
is
an
d
m
ac
h
i
n
e
lear
n
i
n
g
[
2
8
]
,
[2
9
]
.
Giv
en
t
h
e
s
en
s
itiv
ity
o
f
th
e
ch
o
s
en
m
o
d
el
to
o
u
tl
ier
s
,
we
em
p
lo
y
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
3
,
Ma
r
ch
20
25
:
1
588
-
1
5
9
5
1592
th
e
Stan
d
ar
d
-
Scaler
t
o
s
tan
d
a
r
d
ize
o
u
r
d
ataset,
an
d
th
is
h
ad
a
p
o
s
itiv
e
im
p
ac
t
o
n
tr
ai
n
in
g
th
e
m
o
d
el,
as
illu
s
tr
ated
in
T
ab
le
4
.
(
n
o
te:
in
th
is
ex
p
er
ien
ce
,
we
tr
ain
ed
th
e
m
o
d
el
o
n
ly
o
n
o
r
ig
in
al
d
ata
with
o
u
t a
u
g
m
en
ted
d
ata)
.
T
ab
le
4
.
T
h
e
im
p
ac
t o
f
d
ataset
s
tan
d
ar
d
izatio
n
o
n
m
o
d
el
p
er
f
o
r
m
an
ce
Le
a
r
n
i
n
g
r
a
t
e
/
d
a
t
a
s
e
t
W
i
t
h
o
u
t
st
a
n
d
a
r
d
i
z
a
t
i
o
n
%
W
i
t
h
st
a
n
d
a
r
d
i
z
a
t
i
o
n
%
T
r
a
i
n
i
n
g
a
c
c
u
r
a
c
y
7
8
.
9
2
8
2
.
0
0
V
a
l
i
d
a
t
i
o
n
a
c
c
u
r
a
c
y
7
0
.
1
9
8
1
.
5
1
T
r
a
i
n
i
n
g
l
o
ss
5
5
.
7
4
5
1
.
9
7
V
a
l
i
d
a
t
i
o
n
l
o
s
s
6
7
.
3
5
5
7
.
2
6
3
.
2
.
F
e
a
t
ure
s
elec
t
io
n
Fiv
e
f
ea
tu
r
es
wer
e
ev
alu
ate
d
t
o
s
elec
t
th
e
m
o
s
t
s
ig
n
if
ican
t
f
ea
tu
r
es
f
o
r
tr
ain
in
g
o
u
r
m
o
d
el
,
in
clu
d
i
n
g
MFC
C
,
m
el
s
p
ec
tr
o
g
r
am
,
s
p
ec
tr
al
with
its
5
v
ar
ian
ts
,
R
MS,
an
d
Z
C
R
.
Star
tin
g
with
MFC
C
,
a
cr
u
cial
f
ea
tu
r
e,
we
wan
ted
t
o
i
n
v
esti
g
ate
h
o
w
m
an
y
c
o
ef
f
icien
ts
to
in
clu
d
e.
W
h
ile
th
e
f
ir
s
t
1
3
co
ef
f
icien
t
s
ar
e
o
f
ten
s
ee
n
as
th
e
m
o
s
t
r
elev
an
t,
o
u
r
test
s
as
s
h
o
wn
in
T
ab
le
5
,
r
ev
ea
le
d
th
at
o
p
tin
g
f
o
r
4
0
c
o
ef
f
ici
en
ts
led
to
a
b
etter
lear
n
in
g
r
ate.
Hen
ce
,
we
d
ec
id
ed
to
in
cl
u
d
e
4
0
c
o
ef
f
icien
ts
f
r
o
m
th
e
MFC
C
f
ea
tu
r
e.
(
n
o
te:
in
th
is
ex
p
er
ien
ce
,
we
tr
ain
ed
th
e
m
o
d
el
o
n
ly
o
n
o
r
ig
in
al
d
ata
with
o
u
t a
u
g
m
en
t
ed
d
ata)
.
T
ab
le
5
.
T
h
e
im
p
ac
t o
f
MFC
C
d
im
en
s
io
n
n
u
m
b
e
r
o
n
m
o
d
el
p
er
f
o
r
m
an
ce
Le
a
r
n
i
n
g
r
a
t
e
/
d
a
t
a
s
e
t
W
i
t
h
1
3
M
F
C
C
f
e
a
t
u
r
e
s
%
W
i
t
h
4
0
M
F
C
C
f
e
a
t
u
r
e
s
%
Tr
a
i
n
i
n
g
a
c
c
u
r
a
c
y
8
1
.
0
0
8
7
.
8
1
V
a
l
i
d
a
t
i
o
n
a
c
c
u
r
a
c
y
8
1
.
5
1
8
2
.
5
1
Tr
a
i
n
i
n
g
l
o
ss
3
3
.
0
0
0
2
.
0
4
V
a
l
i
d
a
t
i
o
n
l
o
ss
5
2
.
5
4
0
2
.
5
1
Af
ter
s
elec
tin
g
th
e
n
u
m
b
e
r
o
f
co
ef
f
icien
ts
f
o
r
t
h
e
MFC
C
f
ea
tu
r
e,
we
ass
ess
ed
th
e
in
f
lu
e
n
c
e
o
f
o
th
e
r
f
ea
tu
r
es.
T
ab
le
6
p
r
esen
ts
th
e
o
u
tco
m
es
o
f
v
ar
io
u
s
ex
p
e
r
im
en
ts
co
n
d
u
cted
u
s
in
g
in
d
i
v
id
u
al
f
ea
tu
r
es
an
d
co
m
b
in
atio
n
s
.
T
h
er
ef
o
r
e
,
t
h
e
b
est
r
esu
lts
wer
e
o
b
tain
ed
b
y
co
m
b
in
in
g
th
e
f
o
llo
win
g
f
ea
t
u
r
es:
MFC
C
,
R
M
S,
Mel
-
s
p
ec
tr
o
g
r
am
,
a
n
d
s
p
ec
tr
a
l
with
its
5
v
ar
ian
ts
,
r
esu
ltin
g
in
a
tr
ain
in
g
an
d
v
alid
atio
n
a
cc
u
r
ac
y
o
f
9
9
.
5
5
%
an
d
9
9
.
4
2
%
r
esp
ec
tiv
ely
.
T
h
er
ef
o
r
e,
th
e
co
m
b
in
atio
n
o
f
f
ea
tu
r
es
en
h
an
ce
d
th
e
class
if
ier
s
'
p
er
f
o
r
m
an
ce
,
lead
in
g
to
h
ig
h
er
ac
c
u
r
ac
y
co
m
p
ar
ed
to
in
d
iv
id
u
al
f
ea
tu
r
es.
T
ab
le
6.
R
ec
o
g
n
itio
n
r
ate
f
o
r
d
if
f
er
en
t
f
ea
tu
r
e
ex
t
r
ac
tio
n
co
m
b
in
atio
n
s
F
e
a
t
u
r
e
s
V
a
l
i
d
a
t
i
o
n
l
o
ss
%
Tr
a
i
n
i
n
g
a
c
c
u
r
a
c
y
%
V
a
l
i
d
a
t
i
o
n
a
c
c
u
r
a
c
y
%
M
F
C
C
4
0
0
2
.
0
4
9
9
.
3
3
9
9
.
2
2
M
F
C
C
4
0
+
R
M
S
0
1
.
7
5
9
9
.
4
8
9
9
.
0
9
M
F
C
C
4
0
+
R
M
S
+
M
e
l
s
p
e
c
0
1
.
9
5
9
9
.
5
3
9
9
.
0
3
M
F
C
C
4
0
+
R
M
S
+
M
e
l
s
p
e
c
+
sp
e
c
t
r
a
l
0
1
.
3
5
9
9
.
5
5
9
9
.
4
2
M
F
C
C
4
0
+
R
M
S
+
M
e
l
s
p
e
c
+
s
p
e
c
t
r
a
l
+
Z
C
R
0
1
.
3
9
9
9
.
5
6
9
8
.
5
1
T
h
e
ex
p
er
im
en
t
was
co
n
d
u
cte
d
u
s
in
g
th
e
p
r
o
g
r
am
m
in
g
lan
g
u
ag
e
Py
th
o
n
3
.
1
0
.
1
2
,
Ker
as
3
.
0
.
1
,
a
n
d
Go
o
g
le
C
o
lab
,
with
c
o
m
p
u
tatio
n
s
p
er
f
o
r
m
ed
o
n
a
C
PU.
T
h
e
d
u
r
atio
n
o
f
ea
ch
ex
p
e
r
i
m
en
t
r
an
g
ed
f
r
o
m
ap
p
r
o
x
im
ately
3
2
t
o
4
5
m
in
u
t
es.
C
o
n
s
id
er
in
g
t
h
e
class
if
ier
h
y
p
er
p
ar
am
eter
,
th
e
o
p
tim
al
r
esu
lts
wer
e
ac
h
iev
ed
af
ter
2
0
ep
o
ch
s
,
u
tili
zin
g
a
b
a
tch
s
ize
o
f
3
2
,
an
d
im
p
lem
e
n
t
in
g
a
lea
r
n
in
g
r
ate
r
ed
u
ctio
n
t
o
0
.
0
0
0
5
,
we
u
s
e
d
th
e
“Ad
am
”
o
p
tim
izer
an
d
“c
ateg
o
r
ical
cr
o
s
s
-
en
tr
o
p
y
”
as
a
lo
s
s
f
u
n
ctio
n
.
T
h
is
co
n
f
i
g
u
r
ati
o
n
p
lay
e
d
a
c
r
u
cial
r
o
le
in
en
h
an
cin
g
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
.
Fo
r
a
d
ee
p
er
an
aly
s
is
o
f
th
e
r
esu
lts
,
we
co
n
s
tr
u
cted
co
n
f
u
s
io
n
m
atr
ices
illu
s
tr
atin
g
th
e
class
if
ier
s
'
p
er
f
o
r
m
a
n
ce
in
p
r
e
d
ictin
g
ea
ch
em
o
tio
n
as
s
h
o
wn
in
Fig
u
r
e
1
.
On
th
ese
m
atr
ices,
th
e
x
-
ax
is
s
ig
n
if
ies
th
e
p
r
ed
icted
la
b
els,
wh
ile
th
e
y
-
a
x
is
s
ig
n
if
ies
th
e
tr
u
e
lab
els.
N
o
tab
ly
,
t
h
e
an
g
r
y
em
o
tio
n
cla
s
s
ex
h
ib
ited
r
o
b
u
s
t
p
r
ed
ictio
n
s
,
ac
h
iev
in
g
th
e
h
i
g
h
est
ac
cu
r
ac
y
r
ate
o
f
9
9
.
7
3
%.
T
h
is
r
esu
lt
ca
n
b
e
ex
p
lain
e
d
b
y
th
e
f
ac
t
th
at
th
e
an
g
r
y
em
o
tio
n
s
class
co
n
tain
s
h
ig
h
f
r
eq
u
en
cy
an
d
p
itch
as
s
h
o
wn
in
Fig
u
r
e
s
2
a
n
d
3
wh
ich
ar
e
ea
s
ily
ca
p
tu
r
ed
b
y
th
e
C
NN
class
if
ier
.
As
illu
s
tr
ated
in
T
a
b
le
7
,
th
e
r
esu
lts
a
r
e
co
m
p
ar
e
d
to
t
h
e
s
tate
o
f
t
h
e
ar
t,
d
em
o
n
s
tr
atin
g
t
h
at
o
u
r
m
o
d
el
o
u
tp
er
f
o
r
m
ed
ex
is
tin
g
r
esear
c
h
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
C
o
n
vo
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
fo
r
s
p
ee
ch
emo
tio
n
r
ec
o
g
n
itio
n
in
th
e
Mo
r
o
cc
a
n
A
r
a
b
ic
…
(
S
o
u
fiya
n
Ou
a
li
)
1593
Fig
u
r
e
1
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
o
u
r
Ar
ab
ic
SER m
o
d
el
Fig
u
r
e
2
.
W
av
e
p
l
o
t f
o
r
au
d
io
with
h
ap
p
y
em
o
tio
n
Fig
u
r
e
3
.
W
av
e
p
l
o
t f
o
r
au
d
io
with
an
g
r
y
e
m
o
tio
n
T
ab
le
7
.
C
o
m
p
a
r
in
g
o
u
r
m
o
d
e
l w
ith
s
tate
-
of
-
th
e
-
ar
t m
o
d
els
M
o
d
e
l
A
c
c
u
r
a
c
y
[
1
2
]
A
r
a
b
i
c
S
E
R
w
i
t
h
B
A
V
ED
d
a
t
a
se
t
8
9
%
[
1
5
]
A
r
a
b
i
c
(
E
g
y
p
t
i
a
n
d
i
a
l
e
c
t
)
S
E
R
8
8
.
3
%
[
30
]
A
r
a
b
i
c
(
S
a
u
d
i
d
i
a
l
e
c
t
)
S
E
R
7
7
.
1
4
%
[3
1
]
A
r
a
b
i
c
(
A
l
g
e
r
i
a
n
d
i
a
l
e
c
t
)
S
ER
9
3
.
3
4
O
u
r
mo
d
e
l
,
A
r
a
b
i
c
(
M
o
r
o
c
c
a
n
d
i
a
l
e
c
t
)
S
ER
9
9
.
5
5
%
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
3
,
Ma
r
ch
20
25
:
1
588
-
1
5
9
5
1594
4.
CO
NCLU
SI
O
N
W
ith
th
e
ad
v
an
ce
m
en
t
o
f
AI
an
d
th
e
au
t
o
m
atio
n
o
f
v
ar
i
o
u
s
s
ec
to
r
s
,
au
to
m
atic
s
p
ee
ch
r
ec
o
g
n
itio
n
h
as
em
er
g
e
d
as
a
p
r
o
m
in
en
t
r
esear
ch
f
ield
,
o
f
f
er
in
g
alter
n
ativ
es
to
r
o
u
tin
e
h
u
m
an
task
s
in
ar
ea
s
lik
e
c
all
ce
n
ter
s
,
h
ea
lth
ca
r
e
,
v
i
r
tu
al
as
s
is
tan
ts
,
an
d
d
o
m
ai
n
s
s
u
ch
a
s
s
m
ar
t
cities,
s
m
ar
t
h
o
m
es,
an
d
s
m
ar
t
d
ev
ices.
W
h
ile
ex
ten
s
iv
e
r
esear
ch
h
as
b
ee
n
co
n
d
u
cted
in
E
n
g
lis
h
,
th
e
ex
p
lo
r
atio
n
o
f
th
is
f
ield
in
t
h
e
Ar
ab
ic
lan
g
u
ag
e
is
s
t
ill
in
i
ts
ea
r
ly
s
tag
es.
T
h
is
p
ap
er
co
n
tr
ib
u
tes
to
th
e
f
ield
o
f
SR
b
y
d
ev
elo
p
in
g
an
ef
f
ic
ien
t
m
o
d
el
ca
p
ab
le
o
f
r
ec
o
g
n
izin
g
th
e
em
o
tio
n
al
s
tate
o
f
th
e
s
p
ea
k
er
in
th
e
Mo
r
o
cc
an
Ar
a
b
ic
d
ialec
t
lan
g
u
ag
e.
T
h
r
o
u
g
h
a
s
er
ies
o
f
ex
p
er
im
e
n
ts
in
v
o
lv
in
g
d
at
aset
cr
ea
tio
n
,
f
ea
tu
r
e
ex
tr
ac
ti
o
n
,
an
d
clas
s
if
ier
m
o
d
el
s
ele
ctio
n
,
we
id
en
tifie
d
o
p
tim
al
co
m
b
i
n
atio
n
s
to
b
u
ild
an
ef
f
ec
tiv
e
m
o
d
el;
d
ata
a
u
g
m
en
tatio
n
ex
h
ib
ited
a
n
o
tewo
r
th
y
en
h
a
n
ce
m
en
t
in
m
o
d
el
lear
n
in
g
,
r
esu
ltin
g
in
an
1
1
.
5
2
%
in
cr
ea
s
e
in
tr
ai
n
in
g
ac
cu
r
ac
y
.
Stan
d
a
r
d
izin
g
t
h
e
d
ataset
f
u
r
th
e
r
en
h
an
ce
d
m
o
d
el
lear
n
in
g
,
r
es
u
ltin
g
in
a
2
.
0
8
%
in
cr
ea
s
e
in
tr
ain
in
g
ac
cu
r
ac
y
,
w
h
ile
s
elec
tin
g
th
e
ap
p
r
o
p
r
iate
n
u
m
b
er
o
f
MFC
C
co
ef
f
icien
ts
b
o
o
s
ted
tr
ain
in
g
ac
cu
r
ac
y
with
an
in
cr
ea
s
e
o
f
6
.
8
1
%
in
tr
ain
in
g
ac
cu
r
ac
y
.
C
o
m
b
in
in
g
MFC
C
,
m
el
-
s
p
ec
t
r
o
g
r
am
,
s
p
ec
tr
al
f
ea
tu
r
es,
a
n
d
R
MS
w
ith
a
C
N
N
m
o
d
el
f
u
r
th
er
im
p
r
o
v
ed
th
e
m
o
d
el
p
er
f
o
r
m
a
n
ce
,
r
aisi
n
g
t
h
e
ac
cu
r
ac
y
f
r
o
m
9
9
.
3
3
%
to
9
9
.
5
5
%.
T
h
e
m
o
d
el
ac
h
iev
ed
p
r
o
m
is
in
g
r
esu
lts
,
with
a
9
9
.
5
5
%
ac
cu
r
ac
y
r
ate
f
o
r
th
e
s
p
ea
k
er
'
s
em
o
tio
n
al
s
t
ate.
T
h
e
ch
allen
g
e
o
f
lim
ited
d
ata
p
r
o
m
p
te
d
o
u
r
d
ec
i
s
io
n
to
b
u
ild
a
m
o
d
el
p
r
e
d
ictin
g
o
n
ly
f
o
u
r
em
o
tio
n
al
s
tates.
I
n
f
u
tu
r
e
wo
r
k
,
we
aim
to
cr
ea
te
a
lar
g
er
d
ataset
en
co
m
p
ass
in
g
at
least
eig
h
t
em
o
tio
n
al
s
tates
an
d
e
x
p
lo
r
e
ad
d
itio
n
al
p
r
o
s
o
d
ic
an
d
s
p
ec
tr
al
f
ea
tu
r
es
to
f
u
r
th
er
en
h
an
ce
t
h
e
m
o
d
el'
s
ac
cu
r
ac
y
.
Fu
r
th
er
m
o
r
e,
b
u
ild
in
g
a
SER
m
o
d
el
f
o
r
th
e
Mo
r
r
o
ca
n
Ar
ab
ic
d
ialec
t
u
s
in
g
T
r
an
s
f
o
r
m
er
s
as th
ey
h
a
v
e
s
h
o
wn
n
o
tab
le
r
esu
lts
in
th
i
s
f
ield
.
RE
F
E
R
E
NC
E
S
[
1
]
M
.
G
.
H
e
l
a
n
d
e
r
,
“
Em
e
r
g
i
n
g
o
f
f
i
c
e
a
u
t
o
m
a
t
i
o
n
sy
s
t
e
ms
,
”
H
u
m
a
n
Fa
c
t
o
rs
,
v
o
l
.
2
7
,
n
o
.
1
,
p
p
.
3
–
2
0
,
F
e
b
.
1
9
8
5
,
d
o
i
:
1
0
.
1
1
7
7
/
0
0
1
8
7
2
0
8
8
5
0
2
7
0
0
1
0
2
.
[
2
]
B
.
A
.
M
y
e
r
s,
“
A
b
r
i
e
f
h
i
st
o
r
y
o
f
h
u
m
a
n
-
c
o
m
p
u
t
e
r
i
n
t
e
r
a
c
t
i
o
n
t
e
c
h
n
o
l
o
g
y
,
”
I
n
t
e
r
a
c
t
i
o
n
s
,
v
o
l
.
5
,
n
o
.
2
,
p
p
.
4
4
–
5
4
,
M
a
r
.
1
9
9
8
,
d
o
i
:
1
0
.
1
1
4
5
/
2
7
4
4
3
0
.
2
7
4
4
3
6
.
[
3
]
A
.
M
.
G
h
a
n
b
a
r
i
,
S
.
G
h
a
n
b
a
r
i
,
a
n
d
Y
.
N
o
r
o
u
z
i
,
“
A
n
e
w
a
p
p
r
o
a
c
h
t
o
a
r
c
h
i
t
e
c
t
u
r
e
o
f
h
u
m
a
n
-
c
o
mp
u
t
e
r
i
n
t
e
r
a
c
t
i
o
n
,
”
i
n
2
0
1
7
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
S
m
a
rt
I
n
st
r
u
m
e
n
t
a
t
i
o
n
,
Me
a
su
r
e
m
e
n
t
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s,
I
C
S
I
M
A
2
0
1
7
,
N
o
v
.
2
0
1
7
,
v
o
l
.
2
0
1
7
-
N
o
v
e
mb
e
r
,
p
p
.
1
–
4
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
S
I
M
A
.
2
0
1
7
.
8
3
1
1
9
9
1
.
[
4
]
H
.
Tr
a
u
n
m
ü
l
l
e
r
,
“
P
a
r
a
l
i
n
g
u
i
s
t
i
c
P
h
e
n
o
me
n
a
,
”
i
n
S
o
c
i
o
l
i
n
g
u
i
s
t
i
c
s
,
W
a
l
t
e
r
d
e
G
r
u
y
t
e
r
,
2
0
1
7
,
p
p
.
6
5
3
–
6
6
5
.
[
5
]
J
.
H
o
o
k
,
F
.
N
o
r
o
o
z
i
,
O
.
T
o
y
g
a
r
,
a
n
d
G
.
A
n
b
a
r
j
a
f
a
r
i
,
“
A
u
t
o
m
a
t
i
c
s
p
e
e
c
h
b
a
s
e
d
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
u
s
i
n
g
p
a
r
a
l
i
n
g
u
i
s
t
i
c
s
f
e
a
t
u
r
e
s
,
”
B
u
l
l
e
t
i
n
o
f
t
h
e
P
o
l
i
s
h
A
c
a
d
e
m
y
o
f
S
c
i
e
n
c
e
s
:
T
e
c
h
n
i
c
a
l
S
c
i
e
n
c
e
s
,
v
o
l
.
6
7
,
n
o
.
3
,
p
p
.
4
7
9
–
4
8
8
,
2
0
1
9
,
d
o
i
:
1
0
.
2
4
4
2
5
/
b
p
a
s
t
s
.
2
0
1
9
.
1
2
9
6
4
7
.
[
6
]
B
.
M
i
s
h
r
a
,
“
R
o
l
e
o
f
p
a
r
a
l
a
n
g
u
a
g
e
i
n
e
f
f
e
c
t
i
v
e
e
n
g
l
i
sh
c
o
mm
u
n
i
c
a
t
i
o
n
,
”
I
C
F
AI
J
o
u
r
n
a
l
o
f
S
o
f
t
S
k
i
l
l
s
,
v
o
l
.
3
,
n
o
.
2
,
2
0
0
9
.
[
7
]
B
.
W
.
S
c
h
u
l
l
e
r
,
“
S
p
e
e
c
h
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
:
Tw
o
d
e
c
a
d
e
s
i
n
a
n
u
t
s
h
e
l
l
,
b
e
n
c
h
mar
k
s,
a
n
d
o
n
g
o
i
n
g
t
r
e
n
d
s,
”
C
o
m
m
u
n
i
c
a
t
i
o
n
s
o
f
t
h
e
A
C
M
,
v
o
l
.
6
1
,
n
o
.
5
,
p
p
.
9
0
–
9
9
,
A
p
r
.
2
0
1
8
,
d
o
i
:
1
0
.
1
1
4
5
/
3
1
2
9
3
4
0
.
[
8
]
T.
M
.
W
a
n
i
,
T
.
S
.
G
u
n
a
w
a
n
,
S
.
A
.
A
.
Q
a
d
r
i
,
M
.
K
a
r
t
i
w
i
,
a
n
d
E.
A
mb
i
k
a
i
r
a
j
a
h
,
“
A
c
o
m
p
r
e
h
e
n
si
v
e
r
e
v
i
e
w
o
f
s
p
e
e
c
h
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
sy
s
t
e
m
s,”
I
EEE
Ac
c
e
ss
,
v
o
l
.
9
,
p
p
.
4
7
7
9
5
–
4
7
8
1
4
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
1
.
3
0
6
8
0
4
5
.
[
9
]
Y
.
G
a
o
,
B
.
L
i
,
N
.
W
a
n
g
,
a
n
d
T.
Zh
u
,
“
S
p
e
e
c
h
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
u
s
i
n
g
l
o
c
a
l
a
n
d
g
l
o
b
a
l
f
e
a
t
u
r
e
s
,
”
i
n
L
e
c
t
u
re
N
o
t
e
s
i
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
(
i
n
c
l
u
d
i
n
g
s
u
b
s
e
ri
e
s
L
e
c
t
u
r
e
N
o
t
e
s
i
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
L
e
c
t
u
re
N
o
t
e
s
i
n
B
i
o
i
n
f
o
rm
a
t
i
c
s)
,
v
o
l
.
1
0
6
5
4
LN
A
I
,
S
p
r
i
n
g
e
r
I
n
t
e
r
n
a
t
i
o
n
a
l
P
u
b
l
i
sh
i
n
g
,
2
0
1
7
,
p
p
.
3
–
1
3
.
[
1
0
]
Y
.
H
i
f
n
y
a
n
d
A
.
A
l
i
,
“
Ef
f
i
c
i
e
n
t
A
r
a
b
i
c
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
s,”
i
n
I
C
A
S
S
P,
I
E
EE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
A
c
o
u
s
t
i
c
s,
S
p
e
e
c
h
a
n
d
S
i
g
n
a
l
Pr
o
c
e
ss
i
n
g
-
Pr
o
c
e
e
d
i
n
g
s
,
M
a
y
2
0
1
9
,
v
o
l
.
2
0
1
9
-
M
a
y
,
p
p
.
6
7
1
0
–
6
7
1
4
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
S
S
P
.
2
0
1
9
.
8
6
8
3
6
3
2
.
[
1
1
]
A
.
M
e
f
t
a
h
,
Y
.
A
l
o
t
a
i
b
i
,
a
n
d
S
.
A
.
S
e
l
o
u
a
n
i
,
“
D
e
si
g
n
i
n
g
,
b
u
i
l
d
i
n
g
,
a
n
d
a
n
a
l
y
z
i
n
g
a
n
a
r
a
b
i
c
s
p
e
e
c
h
e
m
o
t
i
o
n
a
l
c
o
r
p
u
s
,
”
I
n
W
o
rks
h
o
p
o
n
Fr
e
e
/
O
p
e
n
-
S
o
u
rc
e
Ar
a
b
i
c
C
o
r
p
o
r
a
a
n
d
C
o
rp
o
r
a
Pr
o
c
e
ssi
n
g
T
o
o
l
s
Wo
r
k
sh
o
p
Pr
o
g
r
a
m
m
e
,
v
o
l
.
2
2
,
2
0
1
4
.
[
1
2
]
S
.
A
h
me
d
,
“
A
S
E
R
:
A
r
a
b
i
c
s
p
e
e
c
h
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
e
mp
l
o
y
i
n
g
W
a
v
2
v
e
c
2
.
0
a
n
d
H
u
B
ER
T
b
a
se
d
o
n
B
A
V
ED
d
a
t
a
set
,
”
T
ra
n
s
a
c
t
i
o
n
s
o
n
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
a
n
d
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
9
,
n
o
.
6
,
p
p
.
1
–
8
,
N
o
v
.
2
0
2
1
,
d
o
i
:
1
0
.
1
4
7
3
8
/
t
ml
a
i
.
9
6
.
1
1
0
3
9
.
[
1
3
]
A
.
A
o
u
f
,
“
B
a
si
c
a
r
a
b
i
c
v
o
c
a
l
e
mo
t
i
o
n
s d
a
t
a
s
e
t
,
”
d
o
i
:
1
0
.
3
4
7
4
0
/
K
A
G
G
LE
/
D
S
/
3
4
5
8
2
8
.
[
1
4
]
S
.
K
l
a
y
l
a
t
,
Z.
O
sm
a
n
,
L.
H
a
m
a
n
d
i
,
a
n
d
R
.
Z
a
n
t
o
u
t
,
“
Emo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
i
n
A
r
a
b
i
c
sp
e
e
c
h
,
”
A
n
a
l
o
g
I
n
t
e
g
r
a
t
e
d
C
i
r
c
u
i
t
s
a
n
d
S
i
g
n
a
l
Pr
o
c
e
ssi
n
g
,
9
6
,
p
p
.
3
3
7
-
3
5
1
,
2
0
1
8
[
1
5
]
L.
A
b
d
e
l
-
H
a
m
i
d
,
“
Eg
y
p
t
i
a
n
A
r
a
b
i
c
sp
e
e
c
h
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
p
r
o
so
d
i
c
,
sp
e
c
t
r
a
l
a
n
d
w
a
v
e
l
e
t
f
e
a
t
u
r
e
s,
”
S
p
e
e
c
h
C
o
m
m
u
n
i
c
a
t
i
o
n
,
v
o
l
.
1
2
2
,
p
p
.
1
9
–
3
0
,
S
e
p
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
s
p
e
c
o
m.
2
0
2
0
.
0
4
.
0
0
5
.
[
1
6
]
M
.
B
.
M
u
s
t
a
f
a
,
M
.
A
.
M
.
Y
u
so
o
f
,
Z
.
M
.
D
o
n
,
a
n
d
M
.
M
a
l
e
k
z
a
d
e
h
,
“
S
p
e
e
c
h
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
r
e
s
e
a
r
c
h
:
a
n
a
n
a
l
y
si
s o
f
r
e
sea
r
c
h
f
o
c
u
s,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
S
p
e
e
c
h
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
2
1
,
n
o
.
1
,
p
p
.
1
3
7
–
1
5
6
,
J
a
n
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
7
7
2
-
0
1
8
-
9
4
9
3
-
x.
[1
7
]
S
.
O
u
a
l
i
,
“
M
o
r
r
o
c
a
n
A
r
a
b
i
c
sp
e
e
c
h
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
d
a
t
a
s
e
t
,
”
G
i
t
h
u
b
,
2
0
2
4
.
h
t
t
p
s
:
/
/
g
i
t
h
u
b
.
c
o
m/
S
o
u
f
i
y
a
n
e
O
u
a
l
i
/
M
o
r
r
o
c
a
n
-
A
r
a
b
i
c
-
S
p
e
e
c
h
-
Em
o
t
i
o
n
-
R
e
c
o
g
n
i
t
i
o
n
-
D
a
t
a
s
e
t
.
[1
8
]
B
.
M
c
F
e
e
e
t
a
l
.
,
“
l
i
b
r
o
s
a
:
a
u
d
i
o
a
n
d
mu
si
c
s
i
g
n
a
l
a
n
a
l
y
s
i
s
i
n
p
y
t
h
o
n
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
1
4
t
h
P
y
t
h
o
n
i
n
S
c
i
e
n
c
e
C
o
n
f
e
re
n
c
e
,
2
0
1
5
,
p
p
.
1
8
–
2
4
,
d
o
i
:
1
0
.
2
5
0
8
0
/
maj
o
r
a
-
7
b
9
8
e
3
e
d
-
0
0
3
.
[1
9
]
Z.
K
.
A
b
d
u
l
a
n
d
A
.
K
.
A
l
-
Ta
l
a
b
a
n
i
,
“
M
e
l
f
r
e
q
u
e
n
c
y
c
e
p
st
r
a
l
c
o
e
f
f
i
c
i
e
n
t
a
n
d
i
t
s
a
p
p
l
i
c
a
t
i
o
n
s:
a
r
e
v
i
e
w
,
”
I
EE
E
Ac
c
e
ss
,
v
o
l
.
1
0
,
p
p
.
1
2
2
1
3
6
–
1
2
2
1
5
8
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
2
.
3
2
2
3
4
4
4
.
[
20
]
L.
E.
B
o
u
c
h
e
r
o
n
a
n
d
P
.
L
.
D
e
Le
o
n
,
“
O
n
t
h
e
i
n
v
e
r
si
o
n
o
f
me
l
-
f
r
e
q
u
e
n
c
y
c
e
p
st
r
a
l
c
o
e
f
f
i
c
i
e
n
t
s
f
o
r
s
p
e
e
c
h
e
n
h
a
n
c
e
m
e
n
t
a
p
p
l
i
c
a
t
i
o
n
s,
”
i
n
I
C
S
E
S
’
0
8
-
I
C
S
ES
2
0
0
8
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
S
i
g
n
a
l
s
a
n
d
El
e
c
t
ro
n
i
c
S
y
st
e
m
s,
Pr
o
c
e
e
d
i
n
g
s
,
2
0
0
8
,
p
p
.
4
8
5
–
4
8
8
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
S
ES.2
0
0
8
.
4
6
7
3
4
7
5
.
[2
1
]
A
.
A
l
so
b
h
a
n
i
,
H
.
M
.
A
.
A
l
a
b
b
o
o
d
i
,
a
n
d
H
.
M
a
h
d
i
,
“
S
p
e
e
c
h
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
c
o
n
v
o
l
u
t
i
o
n
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
s,”
J
o
u
r
n
a
l
o
f
Ph
y
s
i
c
s:
C
o
n
f
e
r
e
n
c
e
S
e
ri
e
s
,
v
o
l
.
1
9
7
3
,
n
o
.
1
,
p
.
1
2
1
6
6
,
A
u
g
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
8
8
/
1
7
4
2
-
6
5
9
6
/
1
9
7
3
/
1
/
0
1
2
1
6
6
.
[2
2
]
E.
B
i
s
o
n
g
,
“
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
s
(
C
N
N
)
,
”
i
n
Bu
i
l
d
i
n
g
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
a
n
d
D
e
e
p
L
e
a
r
n
i
n
g
M
o
d
e
l
s
o
n
G
o
o
g
l
e
C
l
o
u
d
Pl
a
t
f
o
rm
,
B
e
r
k
e
l
e
y
,
C
A
:
A
p
r
e
ss,
2
0
1
9
,
p
p
.
4
2
3
–
4
4
1
.
[2
3
]
S
.
S
k
a
n
s
i
,
“
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
s
,
”
i
n
N
a
t
u
r
e
M
e
t
h
o
d
s
,
v
o
l
.
2
0
,
n
o
.
9
,
S
p
r
i
n
g
e
r
I
n
t
e
r
n
a
t
i
o
n
a
l
P
u
b
l
i
s
h
i
n
g
,
2
0
1
8
,
p
p
.
1
2
1
–
1
3
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
C
o
n
vo
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
fo
r
s
p
ee
ch
emo
tio
n
r
ec
o
g
n
itio
n
in
th
e
Mo
r
o
cc
a
n
A
r
a
b
ic
…
(
S
o
u
fiya
n
Ou
a
li
)
1595
[2
4
]
G
.
Li
u
,
W
.
H
e
,
a
n
d
B
.
Ji
n
,
“
F
e
a
t
u
r
e
f
u
si
o
n
o
f
sp
e
e
c
h
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
b
a
se
d
o
n
d
e
e
p
l
e
a
r
n
i
n
g
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
2
0
1
8
6
t
h
I
EEE
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
N
e
t
w
o
r
k
I
n
f
r
a
st
r
u
c
t
u
re
a
n
d
D
i
g
i
t
a
l
C
o
n
t
e
n
t
,
I
C
-
N
I
D
C
2
0
1
8
,
A
u
g
.
2
0
1
8
,
p
p
.
1
9
3
–
1
9
7
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
N
I
D
C
.
2
0
1
8
.
8
5
2
5
7
0
6
.
[2
5
]
T.
M
.
W
a
n
i
,
T.
S
.
G
u
n
a
w
a
n
,
S
.
A
.
A
.
Q
a
d
r
i
,
H
.
M
a
n
so
r
,
M
.
K
a
r
t
i
w
i
,
a
n
d
N
.
I
smai
l
,
“
S
p
e
e
c
h
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
u
s
i
n
g
c
o
n
v
o
l
u
t
i
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
s
a
n
d
d
e
e
p
st
r
i
d
e
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
s,
”
S
e
p
.
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
W
T5
0
4
4
8
.
2
0
2
0
.
9
2
4
3
6
2
2
.
[2
6
]
W
.
Li
m,
D
.
Ja
n
g
,
a
n
d
T.
Le
e
,
“
S
p
e
e
c
h
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
u
s
i
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
a
n
d
r
e
c
u
r
r
e
n
t
n
e
u
r
a
l
n
e
t
w
o
r
k
s,”
i
n
2
0
1
6
Asi
a
-
Pa
c
i
f
i
c
S
i
g
n
a
l
a
n
d
I
n
f
o
rm
a
t
i
o
n
Pr
o
c
e
ssi
n
g
Ass
o
c
i
a
t
i
o
n
A
n
n
u
a
l
S
u
m
m
i
t
a
n
d
C
o
n
f
e
r
e
n
c
e
,
AP
S
I
PA
2
0
1
6
,
D
e
c
.
2
0
1
7
,
p
p
.
1
–
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
P
S
I
P
A
.
2
0
1
6
.
7
8
2
0
6
9
9
.
[2
7
]
L
.
F
e
r
r
e
i
r
a
-
P
a
i
v
a
,
E
.
A
l
f
a
r
o
-
Es
p
i
n
o
z
a
,
V
.
M
.
A
l
m
e
i
d
a
,
L
.
B
.
F
e
l
i
x
,
a
n
d
R
.
V
.
A
.
N
e
v
e
s,
“
A
s
u
r
v
e
y
o
f
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
f
o
r
a
u
d
i
o
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
O
c
t
.
2
0
2
2
,
d
o
i
:
1
0
.
2
0
9
0
6
/
c
b
a
2
0
2
2
/
3
4
6
9
.
[2
8
]
K
.
Ja
j
u
g
a
a
n
d
M
.
W
a
l
e
si
a
k
,
“
S
t
a
n
d
a
r
d
i
s
a
t
i
o
n
o
f
d
a
t
a
se
t
u
n
d
e
r
d
i
f
f
e
r
e
n
t
m
e
a
s
u
r
e
me
n
t
sc
a
l
e
s,”
i
n
C
l
a
ss
i
f
i
c
a
t
i
o
n
a
n
d
I
n
f
o
rm
a
t
i
o
n
Pro
c
e
ssi
n
g
a
t
t
h
e
T
u
r
n
o
f
t
h
e
Mi
l
l
e
n
n
i
u
m
,
S
p
r
i
n
g
e
r
B
e
r
l
i
n
H
e
i
d
e
l
b
e
r
g
,
2
0
0
0
,
p
p
.
1
0
5
–
1
1
2
.
[2
9
]
M
.
S
.
S
h
a
n
k
e
r
,
M
.
Y
.
H
u
,
a
n
d
M
.
S
.
H
u
n
g
,
“
Ef
f
e
c
t
o
f
d
a
t
a
s
t
a
n
d
a
r
d
i
z
a
t
i
o
n
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
t
r
a
i
n
i
n
g
,
”
O
m
e
g
a
,
v
o
l
.
2
4
,
n
o
.
4
,
p
p
.
3
8
5
–
3
9
7
,
A
u
g
.
1
9
9
6
,
d
o
i
:
1
0
.
1
0
1
6
/
0
3
0
5
-
0
4
8
3
(
9
6
)
0
0
0
1
0
-
2.
[
30
]
R
.
H
.
A
l
j
u
h
a
n
i
,
A
.
A
l
sh
u
t
a
y
r
i
,
a
n
d
S
.
A
l
a
h
d
a
l
,
“
A
r
a
b
i
c
s
p
e
e
c
h
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
f
r
o
m
sa
u
d
i
d
i
a
l
e
c
t
c
o
r
p
u
s,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
9
,
p
p
.
1
2
7
0
8
1
–
1
2
7
0
8
5
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
1
.
3
1
1
0
9
9
2
.
[3
1
]
R
.
Y
a
h
i
a
C
h
e
r
i
f
,
A
.
M
o
u
ssa
o
u
i
,
N
.
F
r
a
h
t
a
,
a
n
d
M
.
B
e
r
r
i
m
i
,
“
Ef
f
e
c
t
i
v
e
s
p
e
e
c
h
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
e
s
f
o
r
A
l
g
e
r
i
a
n
d
i
a
l
e
c
t
,
”
i
n
2
0
2
1
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
f
W
o
m
e
n
i
n
D
a
t
a
S
c
i
e
n
c
e
a
t
T
a
i
f
U
n
i
v
e
rsi
t
y
,
W
i
D
S
T
a
i
f
2
0
2
1
,
M
a
r
.
2
0
2
1
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
W
I
D
S
TA
I
F
5
2
2
3
5
.
2
0
2
1
.
9
4
3
0
2
2
4
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
S
o
u
fi
y
a
n
O
u
a
li
is
a
Ph
.
D
.
stu
d
e
n
t
a
t
S
i
d
i
M
o
h
a
m
e
d
Be
n
Ab
d
e
l
lah
F
e
z
,
M
o
ro
c
c
o
.
Ac
h
iev
e
d
h
is
M
a
ste
r'
s
d
e
g
re
e
i
n
Co
m
p
u
ter
S
c
ien
c
e
sp
e
c
ializin
g
in
d
a
ta
sc
ien
c
e
,
Artifi
c
ial
in
telli
g
e
n
c
e
,
a
n
d
n
a
tu
ra
l
lan
g
u
a
g
e
p
ro
c
e
ss
in
g
fro
m
t
h
e
De
p
a
rt
m
e
n
t
o
f
Co
m
p
u
ter
S
c
ie
n
c
e
F
a
c
u
lt
y
o
f
S
c
ien
c
e
,
S
id
i
M
o
h
a
m
e
d
Be
n
A
b
d
e
ll
a
h
F
e
z
,
M
o
ro
c
c
o
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
so
u
fiy
a
n
e
.
o
u
a
li
@u
sm
b
a
.
a
c
.
m
a
.
S
a
id
El
G
a
r
o
u
a
n
i
is
a
n
As
sista
n
t
P
r
o
fe
ss
o
r
a
t
A
b
d
e
lma
lek
E
ss
a
a
d
i
Un
iv
e
rsit
y
,
F
a
c
u
lt
y
o
f
S
c
ien
c
e
s
Teto
u
a
n
si
n
c
e
2
0
1
2
sp
e
c
ializin
g
i
n
t
h
e
field
o
f
Tec
h
n
ica
l
Co
m
m
u
n
ica
ti
o
n
a
n
d
In
f
o
rm
a
ti
o
n
a
n
d
m
o
b
il
e
sy
ste
m
s.
He
wo
rk
s
a
lso
a
s
a
c
o
n
su
lt
a
n
t
fo
r
se
v
e
ra
l
c
o
m
p
a
n
ies
.
He
c
a
n
b
e
c
o
n
t
a
c
ted
a
t
e
m
a
il
:
sa
id
.
e
l
g
a
ro
u
a
n
i@u
sm
b
a
.
a
c
.
m
a
.
Evaluation Warning : The document was created with Spire.PDF for Python.