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.
40
,
No
.
2
,
N
o
v
em
b
e
r
2
0
2
5
,
p
p
.
6
5
4
~
6
6
6
I
SS
N:
2
5
0
2
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/ijeecs.v
40
.i
2
.
pp
654
-
6
6
6
654
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
Cla
ss
ificatio
n of v
o
ice patho
lo
g
ies u
sing
one dimens
io
na
l f
ea
ture
v
ector and
t
wo
di
mens
io
na
l sca
lo
g
ra
m
Ra
nita
K
hu
m
uk
cha
m
1
,
Sh
a
rm
ila
M
eina
m
2
,
K
is
ho
rj
it
No
ng
m
eik
a
pa
m
3
1
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
C
o
m
mu
n
i
c
a
t
i
o
n
En
g
i
n
e
e
r
i
n
g
,
I
n
d
i
a
n
I
n
st
i
t
u
t
e
o
f
I
n
f
o
r
mat
i
o
n
Te
c
h
n
o
l
o
g
y
M
a
n
i
p
u
r
,
I
mp
h
a
l
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
C
o
m
mu
n
i
c
a
t
i
o
n
En
g
i
n
e
e
r
i
n
g
,
M
I
T,
M
a
n
i
p
u
r
U
n
i
v
e
r
si
t
y
,
I
mp
h
a
l
,
I
n
d
i
a
3
D
e
p
a
r
t
me
n
t
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
,
I
n
d
i
a
n
I
n
st
i
t
u
t
e
o
f
I
n
f
o
r
mat
i
o
n
T
e
c
h
n
o
l
o
g
y
M
a
n
i
p
u
r
,
I
mp
h
a
l
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
an
23
,
2
0
2
5
R
ev
is
ed
J
u
l
30
,
2
0
2
5
Acc
ep
ted
Oct
14
,
2
0
2
5
M
o
st
re
se
a
rc
h
wo
rk
f
o
c
u
s
o
n
ly
o
n
b
i
n
a
ry
c
las
sifica
ti
o
n
o
f
v
o
ice
p
a
th
o
l
o
g
ies
su
c
h
a
s
n
o
rm
a
l
a
n
d
p
a
th
o
lo
g
ica
l
c
las
sifica
ti
o
n
.
Ho
we
v
e
r,
th
e
c
u
r
re
n
t
wo
rk
g
iv
e
s
imp
o
r
tan
c
e
to
m
u
lt
icla
ss
c
las
sifica
ti
o
n
to
o
.
Th
e
p
a
p
e
r
c
o
m
p
a
re
s
o
n
e
-
d
ime
n
sio
n
a
l
(1
D)
fe
a
tu
re
v
e
c
to
rs
b
a
se
d
m
a
c
h
in
e
lea
rn
i
n
g
(M
L)
t
e
c
h
n
iq
u
e
s
a
n
d
tw
o
-
d
ime
n
si
o
n
a
l
(2
D)
sc
a
lo
g
ra
m
ima
g
e
b
a
se
d
d
e
e
p
lea
rn
i
n
g
(
DL)
m
o
d
e
l
fo
r
b
in
a
r
y
a
n
d
m
u
lt
icla
ss
c
las
sifica
ti
o
n
o
f
v
o
ice
p
a
th
o
l
o
g
y
.
Th
e
m
u
lt
icla
ss
c
las
sifica
ti
o
n
c
las
sifies
th
e
v
o
ic
e
sig
n
a
l
in
t
o
f
o
u
r
c
a
teg
o
ries
wh
ic
h
a
re
h
e
a
lt
h
y
,
h
y
p
e
rk
in
e
ti
c
d
y
sp
h
o
n
ia,
h
y
p
o
k
i
n
e
ti
c
d
y
sp
h
o
n
ia,
a
n
d
re
fl
u
x
lary
n
g
it
is.
T
h
e
c
u
rre
n
t
wo
r
k
d
e
m
o
n
stra
tes
th
e
e
v
a
lu
a
ti
o
n
o
f
1
D
fe
a
tu
re
v
e
c
to
rs
e
x
trac
ted
fro
m
sp
e
e
c
h
si
g
n
a
l
su
c
h
a
s
M
F
CC
(m
e
l
-
fre
q
u
e
n
c
y
c
e
p
stra
l
c
o
e
fficie
n
t)
a
n
d
p
it
c
h
wit
h
v
a
ri
o
u
s
M
L
tec
h
n
iq
u
e
s
l
ik
e
K
-
n
e
a
re
st
n
e
i
g
h
b
o
r
(KN
N),
Na
ïv
e
Ba
y
e
s,
a
n
d
d
isc
rimin
a
n
t
a
n
a
ly
sis
(DA
).
An
o
th
e
r
tec
h
n
iq
u
e
th
a
t
u
se
s
ti
m
e
-
fre
q
u
e
n
c
y
sc
a
lo
g
r
a
m
s
d
e
riv
e
d
u
si
n
g
th
re
e
d
iffere
n
t
wa
v
e
lets,
i.
e
.
,
a
n
a
ly
t
ica
l
M
o
rlet
(a
m
o
r),
B
u
m
p
,
a
n
d
M
o
rse
,
a
re
u
se
d
f
o
r
train
in
g
a
p
re
-
train
e
d
G
o
o
g
leN
e
t
a
rc
h
it
e
c
tu
re
,
wh
ich
is
a
v
e
r
y
p
o
p
u
lar
D
L
m
o
d
e
l
.
Ex
p
e
rime
n
tal
re
su
l
ts
sh
o
w
th
a
t
2
D
sc
a
lo
g
ra
m
ima
g
e
b
a
se
d
DL
m
o
d
e
l
fo
r
b
in
a
ry
(
9
6
.
0
5
%
)
a
n
d
m
u
lt
icla
ss
(8
9
.
8
%
)
c
las
sifica
ti
o
n
o
f
v
o
ice
p
a
th
o
lo
g
y
g
iv
e
s b
e
tt
e
r
p
e
rfo
rm
a
n
c
e
wh
il
e
c
o
m
p
a
rin
g
wit
h
1
D fea
tu
re
v
e
c
to
rs
b
a
se
d
M
L
tec
h
n
iq
u
e
s.
K
ey
w
o
r
d
s
:
An
aly
tical
Mo
r
let
(
am
o
r
)
B
u
m
p
Dis
cr
im
in
an
t a
n
aly
s
is
K
-
n
ea
r
est n
eig
h
b
o
r
Mo
r
s
e
Naïv
e
B
ay
es
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
:
R
an
ita
Kh
u
m
u
k
ch
am
Dep
ar
tm
en
t o
f
E
lectr
o
n
ics an
d
C
o
m
m
u
n
icatio
n
E
n
g
in
ee
r
i
n
g
I
n
d
ian
I
n
s
titu
te
o
f
I
n
f
o
r
m
atio
n
T
ec
h
n
o
lo
g
y
Ma
n
ip
u
r
I
m
p
h
al
7
9
5
0
0
2
,
I
n
d
ia
E
m
ail: r
an
itak
h
8
9
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Sp
ee
ch
is
th
e
m
o
s
t b
asic f
o
r
m
o
f
ex
p
r
ess
io
n
,
an
d
a
n
y
ch
a
n
g
e
to
th
e
v
o
ca
l c
o
r
d
in
ter
r
u
p
ts
its
s
ea
m
less
f
lo
w.
Vo
ca
l
f
atig
u
e
,
p
r
ess
u
r
e
,
d
y
s
p
h
o
n
ia,
r
o
u
g
h
n
ess
,
g
lo
tt
al
ass
au
lt,
s
o
r
e
th
r
o
at,
an
d
o
th
er
s
y
m
p
to
m
s
ar
e
ex
ac
er
b
ated
b
y
s
p
ee
ch
p
r
o
b
lem
s
.
L
o
n
g
-
ter
m
v
o
ca
l
co
r
d
ab
u
s
e
ca
n
r
esu
lt
in
d
is
ea
s
es
s
u
ch
as
lar
y
n
g
ea
l
m
alig
n
an
cy
,
f
o
ld
in
g
,
p
o
ly
p
,
an
d
n
o
d
u
le.
T
h
e
h
o
a
r
s
en
ess
o
f
o
n
e'
s
v
o
ice
m
ig
h
t
d
ef
in
e
th
ese
co
n
d
itio
n
s
.
Asi
d
e
f
r
o
m
s
elf
-
ab
u
s
e,
a
s
ed
en
tar
y
life
s
ty
le
m
ay
lead
t
o
an
in
cr
ea
s
e
in
v
o
ice
p
r
o
b
lem
s
[
1
]
,
[
2
]
.
Dee
p
lear
n
in
g
(
DL
)
h
as
s
u
r
p
ass
ed
tr
ad
itio
n
al
class
if
ier
s
s
u
ch
as
Naïv
e
B
ay
es,
d
ec
is
io
n
tr
ee
s
,
K
-
n
ea
r
est
n
ei
g
h
b
o
r
(
KNN)
,
an
d
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM)
.
Sin
ce
th
e
last
s
ev
er
al
d
ec
ad
es,
h
an
d
cr
af
ted
s
p
ee
ch
o
r
ac
o
u
s
t
ic
ch
ar
ac
ter
is
tics
h
av
e
b
ee
n
cr
itical
f
o
r
d
etec
ti
n
g
v
o
ice
d
is
o
r
d
e
r
an
d
th
is
ca
n
n
o
t
b
e
o
v
e
r
lo
o
k
e
d
[
3
]
-
[
7
]
.
Fo
r
d
iag
n
o
s
in
g
v
o
ice
p
ath
o
lo
g
y
,
a
wid
e
r
an
g
e
o
f
lo
n
g
an
d
s
h
o
r
t
f
ea
tu
r
e
d
escr
ip
to
r
s
h
av
e
b
ee
n
em
p
lo
y
ed
.
L
o
n
g
-
ter
m
ch
ar
ac
ter
is
tics
h
av
e
b
ee
n
em
p
lo
y
ed
i
n
ce
r
tain
s
ig
n
if
ican
t
r
esear
ch
s
tu
d
ies
[
8
]
-
[
1
1
]
.
W
ah
ed
[
1
2
]
s
u
g
g
ested
a
s
tu
d
y
to
d
ev
elo
p
a
d
etec
to
r
f
o
r
v
o
ca
l
lar
y
n
x
ab
n
o
r
m
alities
b
y
ex
tr
ac
tin
g
a
m
ix
tu
r
e
o
f
v
ar
io
u
s
f
ea
tu
r
e
d
escr
ip
to
r
s
f
r
o
m
a
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:
2502
-
4
7
5
2
C
la
s
s
if
ica
tio
n
o
f v
o
ice
p
a
th
o
l
o
g
ies u
s
in
g
o
n
e
d
imen
s
io
n
a
l f
ea
tu
r
e
ve
cto
r
a
n
d
…
(
R
a
n
ita
K
h
u
mu
kc
h
a
m
)
655
d
is
ea
s
ed
v
o
ice
s
am
p
le.
Or
o
zc
o
-
Ar
r
o
y
av
e
et
a
l.
[
1
3
]
p
r
esen
ted
a
m
eth
o
d
f
o
r
d
iag
n
o
s
in
g
Par
k
in
s
o
n
'
s
illn
ess
,
p
alate
d
y
s
f
u
n
ctio
n
,
a
n
d
v
o
ca
l
f
o
ld
ab
n
o
r
m
alities
,
h
y
p
er
n
asal
lip
.
An
o
t
h
er
s
em
in
al
s
tu
d
y
th
at
u
s
ed
en
tr
o
p
y
to
d
is
tin
g
u
is
h
b
etwe
en
h
ea
lth
y
a
n
d
d
is
ea
s
ed
v
o
ices
was
s
u
g
g
e
s
ted
in
[
1
4
]
.
Ulo
za
et
a
l.
[
1
5
]
d
escr
ib
es
a
m
u
lt
i
-
class
v
o
ice
p
ath
o
lo
g
y
class
if
ier
th
at
em
p
lo
y
s
a
r
ich
f
ea
tu
r
e
v
ec
to
r
g
e
n
er
ated
f
r
o
m
v
ar
ie
d
a
n
d
co
m
m
o
n
s
p
ee
ch
ch
ar
ac
ter
is
tics
.
R
ey
n
o
ld
s
an
d
R
o
s
e
[
1
6
]
r
em
o
v
ed
ch
ar
ac
ter
i
s
tics
f
r
o
m
s
h
o
r
t
r
eg
u
lar
u
tter
a
n
ce
s
u
s
in
g
a
m
el
-
f
r
e
q
u
e
n
c
y
f
i
l
t
e
r
b
a
n
k
.
P
r
a
v
e
n
a
e
t
a
l
.
[1
7
]
u
s
e
d
t
h
e
G
a
u
s
s
i
a
n
m
i
x
t
u
r
e
m
o
d
e
l
(
G
M
M
)
m
o
d
el
t
o
t
r
a
i
n
1
1
d
i
s
t
i
n
c
t
m
e
l
-
f
r
e
q
u
e
n
c
y
c
e
p
s
t
r
al
c
o
e
f
f
i
cie
n
t
(
M
FC
C
)
c
h
a
r
a
ct
e
r
is
t
i
cs
t
o
d
i
s
t
i
n
g
u
is
h
a
n
o
r
m
a
l
v
o
i
c
e
f
r
o
m
a
d
i
s
o
r
d
e
r
e
d
o
n
e
.
T
h
e
m
ac
h
in
e
lea
r
n
in
g
tech
n
iq
u
es is
ap
p
lied
in
m
o
s
t o
f
th
e
m
ed
ical
ap
p
licatio
n
[
1
8
]
.
2.
M
E
T
H
O
D
Fo
r
s
tu
d
y
in
g
th
e
im
p
lem
en
tat
io
n
r
esu
lts
o
f
o
n
e
-
d
im
en
s
io
n
a
l
(
1
D)
a
n
d
2
D
b
ased
m
ac
h
in
e
lear
n
in
g
(
ML
)
an
d
DL
s
y
s
tem
s
,
two
s
ep
ar
ate
wo
r
k
f
l
o
ws
ar
e
p
r
o
p
o
s
ed
in
th
e
cu
r
r
e
n
t
m
eth
o
d
o
lo
g
y
,
as
d
is
cu
s
s
ed
in
th
e
f
o
llo
win
g
two
s
u
b
s
ec
tio
n
s
.
T
h
e
s
u
m
m
ar
ized
ar
ch
itectu
r
es
o
f
Fig
u
r
es
1
an
d
2
ar
e
s
im
ilar
,
ex
ce
p
t,
t
h
e
f
o
r
m
er
d
ea
ls
with
a
b
in
ar
y
class
p
r
ed
i
ctio
n
an
d
th
e
latter
is
a
m
u
lticlas
s
p
r
ed
icto
r
.
I
n
ca
s
e
o
f
1
D
b
i
n
ar
y
class
if
icatio
n
,
th
e
f
ir
s
t
s
tep
is
t
o
c
o
llect
d
ata
wh
ich
is
in
p
u
t
s
p
ee
ch
,
th
e
s
ec
o
n
d
s
tep
g
o
f
o
r
f
ea
tu
r
e
ex
tr
ac
t
io
n
,
th
e
th
ir
d
s
tep
is
f
o
r
ML
wh
ich
is
tr
ain
in
g
an
d
test
in
g
th
e
s
am
p
le
p
r
o
v
id
e
d
an
d
th
e
last
s
tep
u
n
d
er
g
o
es
class
if
icatio
n
o
f
h
ea
lth
y
an
d
p
ath
o
lo
g
ical.
T
h
e
MFC
C
an
d
p
itch
c
h
ar
ac
ter
is
tics
ar
e
ex
tr
ac
te
d
f
r
o
m
th
e
in
p
u
t
s
ig
n
al
as
a
f
ea
tu
r
e
ex
tr
ac
tio
n
.
KNN
,
Naiv
e
B
ay
e
s
,
an
d
d
is
cr
im
in
an
t
a
n
aly
s
is
(
DA)
ar
e
u
s
ed
f
o
r
tr
ain
in
g
an
d
test
in
g
th
e
s
am
p
les.
I
n
2
D
b
in
a
r
y
class
if
icatio
n
,
t
h
e
in
p
u
t
s
p
ee
c
h
is
c
o
n
v
er
te
d
in
to
tim
e
-
f
r
eq
u
e
n
cy
s
ca
lo
g
r
a
m
an
d
g
o
es
f
o
r
DL
u
s
in
g
Go
o
g
leNe
t
an
d
last
ly
cl
ass
if
icatio
n
.
T
h
e
tim
e
f
r
e
q
u
e
n
cy
s
clo
g
r
am
an
d
DL
m
eth
o
d
is
ex
p
lain
ed
in
th
e
later
s
ec
tio
n
.
I
n
ca
s
e
o
f
1
D
m
u
lticlas
s
c
lass
if
icatio
n
,
th
e
f
ir
s
t
s
tep
i
s
to
co
llect
d
ata
wh
ich
is
in
p
u
t
s
p
ee
ch
,
th
e
s
ec
o
n
d
s
tep
g
o
f
o
r
f
ea
t
u
r
e
e
x
tr
ac
tio
n
,
th
e
th
ir
d
s
tep
is
f
o
r
ML
wh
ich
is
tr
ain
in
g
a
n
d
test
in
g
th
e
s
am
p
le
p
r
o
v
id
e
d
an
d
th
e
last
s
tep
u
n
d
er
g
o
es
class
if
icatio
n
o
f
h
ea
lth
y
,
h
y
p
er
k
i
n
etic
d
y
s
p
h
o
n
ia,
h
y
p
o
k
in
etic
d
s
y
p
h
o
n
ia
an
d
lar
y
n
g
itis
.
As
f
ea
tu
r
e
ex
t
r
ac
tio
n
,
MFC
C
an
d
p
itch
ch
a
r
ac
ter
is
tics
ar
e
ex
tr
ac
ted
f
r
o
m
th
e
in
p
u
t
s
ig
n
al.
KNN,
Naiv
e
B
ay
es,
an
d
DA
a
r
e
u
s
ed
f
o
r
ML
wh
ich
ex
p
lain
ed
i
n
later
s
ec
tio
n
.
I
n
2
D
m
u
tliclas
s
class
if
icatio
n
,
th
e
in
p
u
t
s
p
ee
ch
is
co
n
v
er
ted
i
n
to
tim
e
-
f
r
e
q
u
en
cy
s
ca
lo
g
r
am
an
d
g
o
es
f
o
r
DL
an
d
last
ly
class
if
icatio
n
wh
ich
is
ex
p
lain
ed
in
later
s
ec
tio
n
.
2
.
1
.
Da
t
a
s
et
C
esar
i
et
a
l.
[1
9
]
s
u
g
g
ested
a
v
o
ca
l
p
ath
o
lo
g
y
d
ataset,
wh
ich
will
b
e
u
s
ed
i
n
th
is
s
tu
d
y
.
T
h
e
co
llectio
n
co
n
tain
s
1
5
1
d
is
ea
s
ed
an
d
5
5
h
ea
lth
y
s
p
ee
c
h
s
am
p
les,
r
esp
ec
tiv
ely
.
T
h
er
e
ar
e
th
r
ee
ty
p
es
o
f
ab
n
o
r
m
al
v
o
ices: h
y
p
o
k
in
etic
d
y
s
p
h
o
n
ia,
h
y
p
er
k
in
etic
d
y
s
p
h
o
n
ia,
an
d
r
ef
lu
x
la
r
y
n
g
itis
.
All r
ec
o
r
d
in
g
s
f
ea
tu
r
e
a
4
.
7
6
s
ec
o
n
d
s
u
s
tain
ed
‘
a'
v
o
wel
s
o
u
n
d
at
an
8
k
Hz
s
am
p
li
n
g
r
ate.
T
o
av
o
id
o
v
er
f
itti
n
g
,
ea
ch
s
p
ee
ch
s
am
p
le
is
s
p
lit
in
to
1
0
eq
u
al
len
g
th
s
eg
m
en
ts
o
f
0
.
4
7
6
s
ec
o
n
d
d
u
r
atio
n
,
3
,
8
0
8
s
am
p
lin
g
p
o
i
n
ts
,
an
d
an
8
k
Hz
s
am
p
lin
g
f
r
e
q
u
en
c
y
.
Ov
er
f
itti
n
g
o
r
ex
ce
s
s
iv
e
v
ar
ian
ce
m
ig
h
t
lead
to
m
is
lead
in
g
p
o
s
itiv
e
o
u
tco
m
es.
As
in
d
icate
d
in
T
ab
le
1
,
th
is
ar
r
an
g
em
e
n
t
y
ield
ed
1
,
5
1
0
a
n
d
5
5
0
d
is
ea
s
ed
an
d
h
ea
lth
y
s
p
ee
ch
s
am
p
les,
r
esp
ec
tiv
ely
.
T
o
p
r
ev
e
n
t
th
e
i
s
s
u
e
o
f
class
im
b
alan
ce
,
th
e
t
o
tal
n
u
m
b
e
r
o
f
s
am
p
les
th
at
will
b
e
tr
ain
ed
an
d
test
ed
is
5
5
0
f
o
r
ea
ch
class
.
T
h
e
n
u
m
b
er
o
f
s
eg
m
e
n
ted
s
am
p
les
f
o
r
th
e
h
ea
lth
y
class
,
5
5
0
,
i
s
u
s
ed
as
th
e
u
p
p
e
r
lim
it in
th
is
ca
s
e.
T
h
is
b
alan
ce
d
n
o
.
will su
b
s
eq
u
e
n
tly
tak
e
p
ar
t in
tr
ain
in
g
an
d
test
in
g
.
T
h
er
e
ar
e
4
1
,
7
2
,
an
d
3
8
s
a
m
p
les
f
r
o
m
th
e
h
y
p
o
k
in
etic
d
y
s
p
h
o
n
ia,
h
y
p
er
k
in
etic
d
y
s
p
h
o
n
ia
an
d
r
ef
lu
x
lar
y
n
g
itis
ca
teg
o
r
ies,
r
e
s
p
ec
tiv
ely
,
am
o
n
g
th
e
1
5
1
u
n
-
s
eg
m
en
ted
v
o
ice
s
am
p
les.
I
t'
s
also
wo
r
th
n
o
tin
g
th
at
th
ey
'
r
e
all
d
iv
id
ed
i
n
to
te
n
eq
u
al
-
le
n
g
th
s
p
ee
c
h
s
am
p
le
s
.
T
ab
le
2
s
h
o
ws
th
at
th
e
r
e
ar
e
n
o
w
7
2
0
,
4
1
0
,
an
d
3
8
0
s
am
p
les
a
v
ailab
le
f
o
r
ea
c
h
o
f
th
e
t
h
r
ee
class
es.
T
o
p
r
e
v
en
t
th
e
is
s
u
e
o
f
class
im
b
ala
n
ce
,
th
e
n
u
m
b
er
o
f
s
am
p
les
f
o
r
all
f
o
u
r
class
es
is
k
ep
t
at
3
8
0
,
with
r
ef
l
u
x
lar
y
n
g
itis
h
av
in
g
th
e
f
ewe
s
t.
T
h
is
b
alan
ce
d
n
o
.
will
tak
e
p
ar
t in
f
u
tu
r
e
t
r
ain
in
g
a
n
d
test
in
g
.
T
ab
le
1
.
Data
s
et
d
is
tr
ib
u
tio
n
f
o
r
b
in
a
r
y
p
r
ed
ictio
n
C
l
a
s
s
O
r
i
g
i
n
a
l
n
o
.
S
e
g
m
e
n
t
e
d
n
o
.
B
a
l
a
n
c
e
d
n
o
.
H
e
a
l
t
h
y
55
5
5
0
5
5
0
P
a
t
h
o
l
o
g
i
c
a
l
1
5
1
1
5
1
0
5
5
0
T
ab
le
2
.
Data
s
et
d
is
tr
ib
u
tio
n
f
o
r
m
u
lticlas
s
p
r
ed
ictio
n
C
l
a
s
s
O
r
i
g
i
n
a
l
n
o
.
S
e
g
m
e
n
t
e
d
n
o
.
B
a
l
a
n
c
e
d
n
o
.
H
e
a
l
t
h
y
55
5
5
0
3
8
0
H
y
p
o
k
i
n
e
t
i
c
d
y
s
p
h
o
n
i
a
41
4
1
0
3
8
0
H
y
p
e
r
k
i
n
e
t
i
c
d
y
sp
h
o
n
i
a
72
7
2
0
3
8
0
R
e
f
l
u
x
l
a
r
y
n
g
i
t
i
s
38
3
8
0
3
8
0
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
40
,
No
.
2
,
No
v
em
b
er
20
25
:
6
5
4
-
6
6
6
656
2
.
2
.
F
r
a
m
ewo
r
k
f
o
r
cla
s
s
if
ica
t
io
n us
ing
1
D
f
ea
t
ures
a
nd
m
a
chine le
a
rning
T
h
e
wo
r
k
f
lo
w
o
f
th
e
p
r
o
p
o
s
ed
1
D
f
ea
tu
r
es
-
b
ased
ML
ar
ch
itectu
r
e
is
s
h
o
wn
in
Fig
u
r
e
s
1
(
a)
an
d
2
(
a)
.
I
t w
ill co
n
s
is
t o
f
th
r
ee
s
tag
es a
s
ex
p
lain
ed
in
s
u
b
s
ec
tio
n
.
(
a)
(
b
)
Fig
u
r
e
1
.
T
h
e
p
r
o
p
o
s
ed
a
r
ch
it
ec
tu
r
es f
o
r
(
a
)
1
D
an
d
(
b
)
2
D
l
ea
r
n
in
g
m
o
d
els f
o
r
b
in
ar
y
p
r
e
d
ictio
n
2
.
2
.
1
.
Sp
ee
ch
inp
ut
Sp
ee
ch
s
am
p
les
f
r
o
m
eith
e
r
T
ab
le
1
o
r
T
ab
le
2
will
b
e
u
s
ed
d
ep
en
d
i
n
g
o
n
th
e
ty
p
e
o
f
p
r
ed
ictio
n
m
o
d
el
n
ee
d
ed
,
i.e
.
,
b
in
ar
y
o
r
m
u
lticlas
s
.
R
eg
ar
d
less
o
f
p
r
ed
ictio
n
m
o
d
el,
all
s
am
p
les
h
av
e
an
8
k
Hz
s
am
p
lin
g
f
r
eq
u
e
n
cy
an
d
3
,
8
0
8
s
am
p
lin
g
p
o
in
ts
.
2
.
2
.
2
.
F
ea
t
ure
d
escript
o
rs
T
h
e
MFC
C
an
d
p
itch
ch
ar
ac
t
er
is
tics
ar
e
ex
tr
ac
ted
f
r
o
m
th
e
in
p
u
t
s
ig
n
al.
T
h
ese
two
ch
ar
ac
ter
is
tics
ar
e
r
etr
iev
ed
f
r
o
m
a
s
in
g
le
in
p
u
t
v
o
ice
s
am
p
le
a
n
d
c
o
n
ca
ten
ated
in
to
a
s
in
g
le
v
ec
to
r
.
C
o
n
c
aten
ated
v
ec
to
r
s
o
f
th
is
k
in
d
ar
e
cr
ea
te
d
f
o
r
all
tr
a
in
in
g
s
am
p
les.
T
h
ey
will p
ar
ticip
ate
in
tr
ain
in
g
.
MFC
C
is
an
ac
o
u
s
tic
s
ig
n
al
d
escr
ip
tio
n
p
r
ed
icate
d
o
n
th
e
lin
ea
r
co
s
in
e
tr
an
s
f
o
r
m
o
f
a
lo
g
p
o
wer
s
p
ec
tr
u
m
o
n
a
n
o
n
lin
ea
r
m
el
s
ca
le
o
f
f
r
eq
u
en
c
y
[
20
]
.
T
h
e
MFC
C
f
ea
tu
r
es
ar
e
th
e
co
ef
f
icien
ts
th
at
m
ak
e
u
p
th
e
m
el
-
f
r
e
q
u
en
c
y
ce
p
s
tr
u
m
.
T
h
is
f
r
eq
u
e
n
cy
war
p
in
g
im
p
r
o
v
es
th
e
r
e
p
r
esen
tatio
n
o
f
s
o
u
n
d
an
d
s
p
ee
ch
d
ata.
T
h
e
win
d
o
w
len
g
th
is
s
et
at
3
%
o
f
th
e
s
am
p
lin
g
r
ate,
wh
ich
is
2
4
0
.
An
d
th
e
o
v
er
lap
len
g
t
h
is
f
ix
ed
at
2
.
5
%
o
f
th
e
s
am
p
lin
g
r
ate,
w
h
ich
is
2
0
0
.
T
h
e
o
r
ig
in
al
s
am
p
lin
g
r
ate,
i.e
.
,
8
k
Hz
is
u
tili
ze
d
.
Pit
ch
.
T
h
e
f
u
n
d
am
en
tal
f
r
e
q
u
en
cy
o
r
p
itch
o
f
a
v
o
ice
r
elate
s
to
th
e
n
u
m
b
er
o
f
tim
es
th
e
v
o
ca
l
f
o
ld
s
co
m
e
to
g
eth
er
d
u
r
i
n
g
p
h
o
n
at
io
n
p
er
s
ec
o
n
d
.
T
h
e
a
u
to
-
co
r
r
elatio
n
f
u
n
ctio
n
is
u
s
ed
in
tim
e
-
d
o
m
ain
p
itc
h
p
er
io
d
esti
m
ate
m
eth
o
d
s
(
A
C
F).
T
h
e
m
ain
p
r
i
n
cip
le
b
e
h
in
d
co
r
r
elatio
n
-
b
ased
p
itch
tr
ac
k
in
g
is
th
at
th
e
co
r
r
elatio
n
s
ig
n
al
will
h
av
e
a
s
ig
n
if
ican
t
m
a
g
n
itu
d
e
p
ea
k
d
u
r
in
g
th
e
p
itch
p
e
r
io
d
'
s
lag
.
T
h
e
au
to
c
o
r
r
elatio
n
co
m
p
u
tatio
n
is
p
er
f
o
r
m
e
d
d
i
r
ec
tly
o
n
th
e
wa
v
ef
o
r
m
an
d
is
a
s
im
p
le
ca
lcu
latio
n
[
2
1
]
.
Salh
i
et
a
l.
[2
1
]
co
m
p
u
tes th
e
au
t
o
co
r
r
elatio
n
f
u
n
ctio
n
f
o
r
a
s
ig
n
al
x
(
n
)
.
(
)
=
l
im
→
∞
1
2
+
1
∑
(
)
(
+
)
=
−
(
1
)
T
h
e
au
to
co
r
r
elatio
n
f
u
n
ctio
n
o
f
a
s
ig
n
al
is
b
asically
a
tr
an
s
f
o
r
m
atio
n
o
f
th
e
s
ig
n
al
wh
ich
is
u
s
ef
u
l f
o
r
d
is
p
lay
in
g
s
tr
u
ct
u
r
e
i
n
th
e
wa
v
ef
o
r
m
.
T
h
u
s
,
f
o
r
p
itch
d
etec
tio
n
,
if
we
ass
u
m
e
x
(
n
)
is
e
x
ac
tly
p
er
io
d
ic
with
p
er
io
d
P,
i.
e.
x
(
n
)
=x
(
n
+P)
f
o
r
all
n
,
th
en
th
e
au
to
c
o
r
r
elatio
n
f
u
n
ctio
n
o
f
(
1
)
is
also
p
er
io
d
ic
with
th
e
s
am
e
p
er
io
d
.
(
)
=
(
+
)
(
2
)
2
.
2
.
3
.
M
a
chine
l
ea
rning
cla
s
s
if
iers
T
h
er
e
ar
e
n
u
m
e
r
o
u
s
class
if
icatio
n
alg
o
r
ith
m
s
av
ailab
le
to
d
ay
,
b
u
t
n
o
n
e
o
f
th
em
o
u
tp
e
r
f
o
r
m
t
h
e
o
th
er
s
in
ev
er
y
ca
s
e
[
2
2
]
.
W
e
ch
o
s
e
th
r
ee
class
if
ier
s
f
o
r
th
e
cu
r
r
en
t
wo
r
k
s
tu
d
y
:
K
NN,
Nai
v
e
B
ay
es,
an
d
DA
.
T
h
ese
class
if
ier
s
ar
e
tr
ain
e
d
i
n
d
iv
id
u
ally
u
s
in
g
th
e
co
n
ca
ten
ated
f
ea
t
u
r
e
v
ec
to
r
s
o
b
tain
e
d
f
r
o
m
th
e
t
r
ain
in
g
s
am
p
les.
Ak
b
u
lu
t
et
a
l.
[2
3
]
s
tates
,
th
e
KNN
tech
n
iq
u
e
is
am
o
n
g
th
e
ea
r
lies
t
an
d
ea
s
iest
k
in
d
s
o
f
n
o
n
p
a
r
am
etr
ic
class
if
ier
.
T
h
e
d
r
awb
ac
k
is
th
at
wh
en
a
lo
w
k
v
alu
e
is
u
s
ed
,
t
h
e
s
ep
ar
ati
o
n
b
o
r
d
er
b
ec
o
m
es
ex
ce
s
s
iv
ely
ad
ap
ted
to
th
e
tr
ain
in
g
d
ata,
r
esu
ltin
g
in
o
v
er
-
tr
ain
in
g
.
At
h
ig
h
er
k
v
alu
es,
th
e
b
o
r
d
er
ten
d
s
to
b
e
s
m
o
o
th
er
,
r
esu
ltin
g
in
im
p
r
o
v
ed
p
r
ed
ictio
n
r
esu
lts
f
o
r
f
r
e
s
h
s
am
p
les.
T
h
e
b
est
v
alu
e
o
f
k
m
u
s
t
b
e
f
o
u
n
d
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:
2502
-
4
7
5
2
C
la
s
s
if
ica
tio
n
o
f v
o
ice
p
a
th
o
l
o
g
ies u
s
in
g
o
n
e
d
imen
s
io
n
a
l f
ea
tu
r
e
ve
cto
r
a
n
d
…
(
R
a
n
ita
K
h
u
mu
kc
h
a
m
)
657
em
p
ir
ically
.
T
o
id
e
n
tify
t
h
e
o
p
tim
al
v
alu
e
o
f
k
,
we
e
m
p
i
r
ically
ev
alu
ated
d
if
f
er
en
t
v
a
lu
es
o
f
k
u
s
in
g
th
e
E
u
clid
ea
n
d
is
tan
ce
m
et
r
ic.
Mo
r
e
s
p
ec
if
ically
,
we
test
ed
k
=1
,
2
,
3
,
4
,
5
,
6
,
7
.
I
t
was
d
is
co
v
er
ed
th
at
a
v
alu
e
o
f
k
=5
p
r
o
d
u
ce
s
th
e
g
r
ea
test
r
esu
lts
.
Naiv
e
B
ay
es:
th
e
f
u
n
d
am
en
tal
f
ea
tu
r
e
o
f
Naiv
e
B
ay
es
is
a
s
t
r
o
n
g
n
aiv
e
ass
u
m
p
tio
n
o
f
in
d
e
p
en
d
en
ce
f
r
o
m
ea
ch
co
n
d
itio
n
o
r
o
cc
u
r
r
en
ce
.
I
t
is
a
s
tr
aig
h
tf
o
r
war
d
m
o
d
el
th
at
m
ay
b
e
u
s
ed
to
h
u
g
e
d
atasets
.
T
h
e
b
asis
o
f
th
e
Naiv
e
B
ay
es th
eo
r
em
is
th
e
B
ay
es f
o
r
m
u
la,
wh
ich
is
g
iv
en
b
y
(
|
)
=
(
)
(
|
)
(
)
(
3
)
wh
er
e,
X=
(
x
1
,
x
2
,
x
3
,
…,
x
n
)
is
th
e
attr
ib
u
te,
C
is
th
e
cl
ass
,
P(C
|
X)
:
p
r
o
b
ab
ilit
y
o
f
e
v
en
t
g
iv
en
h
as
o
cc
u
r
r
e
d
,
P(X
|
C
)
: p
r
o
b
ab
ilit
y
o
f
ev
en
t
g
iv
en
h
as o
cc
u
r
r
ed
,
P(C):
p
r
o
b
ab
ilit
y
o
f
ev
en
t
C
,
P(X
)
: p
r
o
b
ab
ilit
y
o
f
ev
en
t X
.
W
e
m
u
s
t
m
ax
im
is
e
th
e
p
r
o
b
ab
ilit
y
v
alu
e
o
f
ea
ch
cla
s
s
in
th
e
Nav
e
B
ay
es
c
lass
if
ier
,
wh
ich
is
r
ep
r
esen
ted
as th
e
h
y
p
o
th
esis
m
ax
im
u
m
a
p
o
s
ter
io
r
i
(
HM
A
P).
=
a
r
g
ma
x
(
|
1
,
2
,
…
…
…
,
)
=
a
r
g
ma
x
(
)
∏
=
1
(
|
)
(
4
)
W
h
er
e,
P r
ep
r
esen
ts
o
p
p
o
r
tu
n
i
ty
,
x
i
is
th
e
i
th
attr
ib
u
te
v
alu
e,
C
is
clas
s
.
L
in
ea
r
d
is
cr
im
in
an
t
a
n
aly
s
is
(
L
DA)
:
it
ca
n
b
e
u
s
ed
f
o
r
c
lass
if
icatio
n
as
well
as
d
im
en
s
io
n
ality
r
ed
u
ctio
n
.
T
h
is
class
if
ier
ev
alu
ates
a
p
r
o
jectio
n
h
y
p
er
p
l
an
e
th
at
ac
co
m
p
lis
h
es
two
g
o
als:
1
)
in
ter
class
v
ar
ian
ce
s
h
o
u
ld
b
e
r
e
d
u
ce
d
,
an
d
2
)
p
r
o
jecte
d
m
ea
n
s
o
f
class
es
s
h
o
u
ld
b
e
as
clo
s
e
to
ea
c
h
o
th
er
as
p
o
s
s
ib
le
[
4
]
.
C
o
n
s
id
er
th
e
f
o
llo
win
g
ex
am
p
le
in
w
h
ich
a
class
is
to
b
e
p
r
ed
icted
.
L
et
X
r
e
p
r
esen
t
th
e
p
r
ed
icto
r
v
ar
iab
les.
Su
p
p
o
s
e
X
is
th
e
s
in
g
le
p
r
e
d
icto
r
v
a
r
iab
le,
i.e
.
X=
x
.
L
et
f
k
(
x
)
b
e
th
e
esti
m
ated
d
is
cr
im
in
ato
r
s
co
r
e
th
at
th
e
o
b
s
er
v
atio
n
b
elo
n
g
s
to
th
e
C
k
class
.
T
h
en
,
f
k
(
x
)
ca
n
b
e
ev
alu
ated
b
y
th
e
f
o
r
m
u
la:
(
)
=
2
−
2
2
2
+
l
og
(
∏
)
(
5
)
w
h
er
e,
∏
is
th
e
p
r
io
r
p
r
o
b
a
b
ilit
y
th
at
th
e
class
o
f
o
b
s
er
v
ati
o
n
i
s
C
k
.
is
t
h
e
av
e
r
ag
e
o
f
tr
ain
in
g
o
b
s
er
v
atio
n
s
b
elo
n
g
in
g
to
class
C
k
.
Fo
r
ea
c
h
o
f
th
e
K
class
es
th
e
weig
h
te
d
av
e
r
ag
e
o
f
s
am
p
le
v
a
r
ian
ce
s
is
r
ep
r
esen
te
d
b
y
2
.
T
h
e
L
DA
class
if
ier
will p
r
ed
ict
th
at
class
k
f
o
r
th
e
g
i
v
en
o
b
s
er
v
atio
n
wh
o
s
e
d
is
cr
im
in
an
t
s
co
r
e
is
lar
g
est.
2
.
3
.
F
r
a
m
ewo
r
k
f
o
r
cla
s
s
if
ica
t
io
n us
ing
2
D
s
ca
lo
g
ra
m
s
a
nd
deep
lea
rni
ng
T
h
e
cu
r
r
en
t
s
u
b
s
ec
tio
n
will
d
is
cu
s
s
th
e
ef
f
ec
ts
o
f
u
s
in
g
an
im
ag
e
-
b
ased
an
aly
s
is
f
o
r
p
er
f
o
r
m
in
g
b
o
t
h
b
in
ar
y
an
d
m
u
lticlas
s
p
r
ed
ictio
n
s
.
T
h
e
wo
r
k
f
lo
w
is
h
ig
h
lig
h
ted
in
Fig
u
r
e
s
1
(
b
)
a
n
d
2
(
b
)
.
T
h
e
f
ir
s
t
s
tep
is
to
g
en
er
ate
s
ca
lo
g
r
am
im
ag
es f
r
o
m
all
s
am
p
les o
f
ev
er
y
class
.
(
a)
(
b
)
Fig
u
r
e
2
.
T
h
e
p
r
o
p
o
s
ed
a
r
ch
it
ec
tu
r
es f
o
r
(
a
)
1
D
an
d
(
b
)
2
D
l
ea
r
n
in
g
m
o
d
els f
o
r
m
u
lticlas
s
p
r
ed
ictio
n
T
im
e
-
f
r
eq
u
en
cy
s
ca
lo
g
r
am
s
:
th
e
n
ex
t
s
tep
is
to
co
n
v
er
t
th
e
s
eg
m
en
ted
s
p
ee
ch
s
am
p
les
f
r
o
m
ea
ch
o
f
th
ese
th
r
ee
class
es
in
to
M
o
r
s
e
s
ca
lo
g
r
am
(
M.
S)
2
D
im
ag
e
s
.
T
h
e
c
o
n
tin
u
o
u
s
wav
elet
tr
a
n
s
f
o
r
m
(
C
W
T
)
o
f
a
g
iv
en
s
ig
n
al
h
a
v
in
g
f
u
n
ctio
n
f
(
t)
is
ev
alu
ated
b
y
u
s
in
g
t
h
e
m
o
th
er
wav
elet
th
r
o
u
g
h
th
e
e
x
p
r
ess
io
n
:
(
,
)
=
1
√
∫
(
)
∗
(
−
)
+
−
(
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
40
,
No
.
2
,
No
v
em
b
er
20
25
:
6
5
4
-
6
6
6
658
w
h
er
e,
x
an
d
y
ar
e
th
e
s
ca
lin
g
an
d
s
h
if
tin
g
f
ac
to
r
f
o
r
th
e
m
o
th
er
wav
elet
an
d
*
s
ig
n
if
ies
co
n
v
o
lu
tio
n
s
o
p
er
atio
n
.
T
h
e
ab
o
v
e
ex
p
r
ess
io
n
ca
n
b
e
tr
a
n
s
lated
as
an
i
n
teg
r
atio
n
o
f
s
u
m
m
ati
o
n
o
f
th
e
in
p
u
t
au
d
io
s
am
p
le
m
u
ltip
lied
b
y
t
h
e
tim
e
s
ca
led
an
d
s
h
if
ted
f
o
r
m
s
o
f
th
e
m
o
th
er
wav
elet
(
m
)
.
T
h
e
Mo
r
s
e
wav
elet
is
b
ein
g
ch
o
s
en
f
o
r
th
e
c
u
r
r
en
t
wo
r
k
b
ec
au
s
e
it
d
is
p
lay
s
s
tr
o
n
g
lo
ca
lizatio
n
in
b
o
th
th
e
f
r
eq
u
en
c
y
an
d
tem
p
o
r
al
d
o
m
ain
s
,
m
ak
in
g
it
id
ea
l
f
o
r
s
tu
d
y
in
g
l
o
ca
lized
d
is
co
n
tin
u
ities
.
T
h
e
f
o
u
r
ier
tr
an
s
f
o
r
m
o
f
a
Mo
r
s
e
wav
elet
is
ex
p
r
ess
ed
as
:
,
(
)
=
(
)
,
2
−
(
7
)
w
h
er
e,
ξ(
)
is
a
u
n
it st
ep
f
u
n
cti
o
n
,
2
is
th
e
tim
e
-
b
an
d
wid
th
p
r
o
d
u
ct,
,
ɳ
s
ig
n
if
ies n
o
r
m
aliza
tio
n
co
n
s
tan
t
an
d
is
th
e
s
y
m
m
etr
y
p
ar
a
m
eter
.
Dif
f
er
en
t c
o
m
b
in
atio
n
o
f
2
d
an
d
ca
n
p
r
o
d
u
ce
d
iv
e
r
s
e
Mo
r
s
e
wav
elets.
Similar
ly
,
th
e
co
ef
f
icien
ts
o
f
(
6
)
ca
n
b
e
im
p
le
m
en
ted
wit
h
b
u
m
p
wav
elet
tr
an
s
f
o
r
m
at
i
o
n
[
2
4
]
to
d
er
iv
e
th
e
g
lo
ttal d
er
iv
ativ
e
B
u
m
p
s
ca
lo
g
r
a
m
.
T
h
e
f
o
u
r
ier
tr
an
s
f
o
r
m
o
f
a
b
u
m
p
wav
elet
is
:
(
)
=
(
1
−
1
1
−
(
−
)
2
2
)
1
[
−
,
+
]
(
8
)
wh
er
e,
an
d
ar
e
p
ar
am
ete
r
s
th
at
co
n
tr
o
ls
th
e
tr
an
s
f
o
r
m
ed
s
ig
n
a
l’
s
f
r
eq
u
en
c
y
an
d
tim
e
lo
ca
liz
atio
n
.
Ap
p
ly
in
g
tim
e
-
d
o
m
ai
n
to
f
r
e
q
u
en
cy
-
d
o
m
ai
n
tr
an
s
f
o
r
m
atio
n
u
s
in
g
wa
v
elet,
th
e
1
-
D
in
p
u
t
s
ig
n
al
is
tr
an
s
f
o
r
m
ed
in
to
a
2
D
s
ig
n
al.
An
d
an
an
al
y
tical
m
o
r
let
(
am
o
r
)
wa
v
elet
b
as
ed
tim
e
-
f
r
eq
u
en
cy
v
er
s
io
n
o
f
th
e
in
p
u
t a
u
d
io
is
:
=
2
−
4
ln
(
2
)
2
ℎ
2
(
9
)
wh
er
e
h
is
f
u
ll
-
wid
th
at
h
alf
-
m
ax
im
u
m
(
FW
HM
)
wh
ich
is
th
e
d
is
tan
ce
in
tim
e
b
etwe
en
5
0
%
g
ain
b
e
f
o
r
e
t
h
e
p
ea
k
to
5
0
% g
ain
a
f
ter
th
e
p
ea
k
[
2
3
]
.
2
.
3
.
1
.
G
o
o
g
L
eNe
t
I
t
is
a
cu
tt
in
g
-
ed
g
e
co
n
v
o
lu
ti
o
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN)
s
u
g
g
ested
b
y
Go
o
g
le.
I
t
h
ad
a
to
p
-
f
iv
e
m
is
tak
e
r
ate
o
f
6
.
6
7
%
[
2
5
]
.
T
h
e
Go
o
g
leNe
t
em
p
lo
y
s
n
i
n
e
(
9
)
1
D
-
in
ce
p
tio
n
m
o
d
u
les,
ea
ch
o
f
w
h
ich
em
p
l
o
y
s
th
r
ee
d
is
tin
ct
c
o
n
v
o
lu
tio
n
al
k
er
n
els,
n
am
ely
1
x
1
,
3
x
3
,
an
d
5
x
5
.
T
h
is
n
etwo
r
k
h
as
a
to
tal
o
f
1
4
2
la
y
er
s
.
T
h
e
in
p
u
t
lay
er
is
a
2
D
im
ag
e
in
p
u
t
lay
er
with
2
2
4
x
2
2
4
x
3
d
im
e
n
s
io
n
s
.
I
t
is
lin
k
ed
to
a
co
n
v
o
lu
tio
n
al
lay
er
with
a
k
er
n
el
s
ize
o
f
7
x
7
,
s
tr
id
e
o
f
2
,
an
d
5
1
2
f
ilter
s
.
T
h
is
lay
er
w
ill
co
llect
f
ea
tu
r
es
f
r
o
m
th
e
p
r
ec
ed
in
g
lay
er
(
th
e
in
p
u
t
lay
er
)
a
n
d
s
to
r
e
th
em
as
ac
tiv
atio
n
m
ap
s
with
5
1
2
d
e
p
t
h
s
(
eq
u
al
to
th
e
n
u
m
b
er
o
f
f
i
lter
s
)
.
I
t
is
lin
k
ed
to
a
m
ax
-
p
o
o
lin
g
lay
er
with
k
er
n
el
s
ize
o
r
f
ilter
s
ize
3
x
3
an
d
s
tr
id
e
eq
u
al
to
2
th
r
o
u
g
h
a
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U
)
lay
er
.
T
h
e
m
ax
-
p
o
o
li
n
g
lay
er
'
s
g
o
al
is
to
d
o
wn
s
am
p
le
(
o
r
m
in
im
is
e)
th
e
s
ize
o
f
th
e
ac
tiv
atio
n
m
ap
s
cr
ea
ted
b
y
th
e
p
r
ev
i
o
u
s
lay
er
.
T
o
m
in
im
is
e
o
v
er
f
itti
n
g
,
th
is
n
ew
ac
tiv
atio
n
m
a
p
is
n
o
w
p
u
t
in
to
a
n
o
r
m
alis
in
g
lay
er
.
Ov
e
r
f
itti
n
g
is
a
p
h
e
n
o
m
en
o
n
t
h
at
r
e
d
u
ce
s
DL
n
etwo
r
k
ac
cu
r
ac
y
b
y
s
u
p
p
ly
in
g
f
ea
tu
r
es
in
a
n
o
n
-
u
n
if
o
r
m
m
an
n
er
.
Ov
e
r
f
itti
n
g
is
m
in
im
i
s
ed
b
y
u
tili
s
in
g
eith
er
a
d
r
o
p
o
u
t
lay
er
o
r
a
n
o
r
m
alis
in
g
lay
er
;
cu
r
r
en
tly
,
d
r
o
p
o
u
t
is
s
eld
o
m
em
p
lo
y
ed
,
an
d
b
at
ch
n
o
r
m
alis
atio
n
o
r
c
r
o
s
s
ch
an
n
el
n
o
r
m
alis
atio
n
h
as
lar
g
ely
r
ep
lace
d
it.
T
h
e
n
o
r
m
alis
ed
lay
er
is
lin
k
ed
to
two
f
u
r
th
er
co
n
v
o
lu
tio
n
al
lay
er
s
with
k
er
n
el
s
izes
o
f
3
x
3
,
s
tr
id
e
2
th
r
o
u
g
h
a
R
eL
U
lay
er
.
W
ith
th
is
s
ec
o
n
d
co
n
v
o
l
u
tio
n
al
la
y
er
,
a
cr
o
s
s
ch
an
n
el
n
o
r
m
alis
atio
n
la
y
er
is
e
m
p
lo
y
ed
,
f
o
llo
wed
b
y
a
m
ax
-
p
o
o
lin
g
lay
er
.
T
h
is
m
ax
-
p
o
o
lin
g
lay
e
r
'
s
ac
tiv
atio
n
m
ap
s
ar
e
lin
k
ed
t
o
an
in
ce
p
tio
n
m
o
d
u
le.
E
ac
h
in
ce
p
tio
n
m
o
d
u
le
in
c
lu
d
es
1
3
lay
er
s
,
6
o
f
wh
ich
ar
e
co
n
v
o
l
u
tio
n
al
lay
er
s
an
d
th
e
r
est
ar
e
a
m
ix
o
f
R
eL
U
an
d
m
ax
-
p
o
o
lin
g
lay
e
r
s
.
A
d
ep
t
h
co
n
ca
ten
atio
n
m
o
d
u
le
is
u
tili
s
ed
at
th
e
co
n
clu
s
io
n
o
f
ea
ch
i
n
ce
p
tio
n
m
o
d
u
le
t
o
m
er
g
e
th
e
ac
tiv
atio
n
m
ap
s
f
r
o
m
th
e
in
ce
p
tio
n
m
o
d
u
le'
s
f
o
u
r
co
lu
m
n
s
.
T
h
e
Go
o
g
L
eNe
t'
s
f
in
al
lay
er
s
in
clu
d
e
d
r
o
p
o
u
t,
f
u
lly
co
n
n
ec
ted
,
s
o
f
tm
ax
,
an
d
a
class
if
icatio
n
o
u
t
p
u
t
lay
er
.
T
h
e
d
r
o
p
o
u
t
lay
e
r
em
p
lo
y
s
a
d
r
o
p
o
u
t
p
r
o
b
a
b
ilit
y
o
f
7
0
%.
T
h
e
d
im
e
n
s
io
n
o
f
th
e
c
o
m
p
letely
lin
k
e
d
lay
er
is
2
,
0
4
8
.
T
h
e
r
elate
d
p
r
o
b
a
b
ilit
ies
will
b
e
co
m
p
u
ted
u
s
in
g
th
e
s
o
f
tm
ax
l
ay
er
.
T
h
e
last
lay
er
is
a
class
i
f
icatio
n
o
u
t
p
u
t
lay
er
,
wh
ich
w
ill
b
e
p
r
o
g
r
am
m
ed
to
id
en
tify
th
e
n
u
m
b
er
o
f
class
es r
eq
u
ested
.
3.
RE
SU
L
T
S
AND
D
I
SCU
SS
I
O
NS
Af
ter
ca
r
ef
u
lly
im
p
lem
e
n
tin
g
th
e
p
r
ec
o
d
u
r
e
in
th
e
m
o
d
el,
t
h
e
ex
p
er
im
e
n
tal
r
esu
lts
ar
e
ev
alu
ated
as
f
o
llo
ws:
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:
2502
-
4
7
5
2
C
la
s
s
if
ica
tio
n
o
f v
o
ice
p
a
th
o
l
o
g
ies u
s
in
g
o
n
e
d
imen
s
io
n
a
l f
ea
tu
r
e
ve
cto
r
a
n
d
…
(
R
a
n
ita
K
h
u
mu
kc
h
a
m
)
659
3
.
1
.
E
v
a
lua
t
i
o
n m
et
rics
T
h
e
cu
r
r
e
n
t
wo
r
k
will
b
e
ev
al
u
ated
u
s
in
g
n
in
e
(
9
)
m
etr
ics,
wh
ich
ar
e
–
s
en
s
itiv
ity
(
Sen
.
)
,
ac
cu
r
ac
y
(
Acc
.
)
,
C
o
h
en
’
s
k
ap
p
a
i
n
d
ex
e
r
r
o
r
(
E
r
r
.
)
,
p
r
ec
is
io
n
(
Pre
.
)
,
s
p
ec
if
icity
(
Sp
e.
)
,
Ma
tth
ews
co
r
r
ela
tio
n
co
ef
f
icien
t
(
MCC
)
,
f
alse
p
o
s
itiv
e
r
ate
(
FP
R
)
,
an
d
F1
s
co
r
e.
Her
e,
T
P,
T
N,
FP
,
F
N
s
tan
d
s
f
o
r
tr
u
e
p
o
s
itiv
e,
tr
u
e
n
eg
ativ
e,
f
alse p
o
s
itiv
e,
an
d
f
alse n
e
g
ativ
e
r
esp
ec
tiv
ely
.
−
Sen
s
itiv
ity
:
i
t
id
en
tifie
s
th
e
ac
tu
al
n
u
m
b
er
o
f
p
o
s
itiv
e
s
am
p
le
s
o
f
all
th
e
p
o
s
i
tiv
es
s
am
p
les.
I
t
is
also
ca
lled
as tr
u
e
p
o
s
itiv
e
r
ate
(
T
PR
)
an
d
is
g
iv
en
b
y
:
=
+
(
1
0
)
−
Acc
u
r
ac
y
:
i
t
is
th
e
s
im
p
lest
an
d
m
o
s
t
co
m
m
o
n
m
et
r
ic
f
o
r
m
o
d
el
ev
al
u
atio
n
.
I
t
is
th
e
r
at
io
o
f
th
e
c
o
r
r
ec
t
p
r
ed
ictio
n
w
h
ich
is
th
e
s
u
m
o
f
T
P
an
d
T
N
to
th
e
to
tal
n
u
m
b
er
o
f
p
r
ed
ictio
n
s
o
f
th
e
g
iv
en
d
ataset
o
r
s
am
p
les wh
ich
is
g
iv
en
b
y
:
=
+
+
+
+
(
1
1
)
−
C
o
h
en
’
s
Kap
p
a
in
d
e
x
:
i
t
is
u
s
ed
to
m
ea
s
u
r
e
th
e
f
e
d
ilit
y
o
f
t
wo
r
ater
s
.
I
f
th
e
v
alu
e
is
less
,
th
en
ze
r
o
th
an
th
er
e
is
n
o
ag
r
ee
m
en
t a
n
d
if
it
is
in
b
etwe
en
0
.
8
1
to
1
th
an
th
er
is
p
er
f
ec
t a
g
r
ee
m
en
t.
=
−
1
−
=
1
−
1
−
1
−
(
1
2
)
−
E
r
r
o
r
:
i
t d
eter
m
in
es th
e
wr
o
n
g
class
if
icatio
n
wh
ich
is
g
iv
en
b
y
(
1
3
)
.
=
100
−
(
1
3
)
−
Pre
cisi
o
n
:
i
t is th
e
r
atio
o
f
th
e
tr
u
e
p
o
s
itiv
es to
all
th
e
p
o
s
itiv
es o
f
th
e
s
am
p
les.
=
+
(
1
4
)
−
Sp
ec
if
icity
:
i
t
id
en
tifie
s
th
e
a
ctu
al
n
u
m
b
er
o
f
n
eg
ativ
e
s
am
p
les
o
f
all
th
e
n
eg
ativ
e
s
a
m
p
les.
Her
e,
it
is
m
o
r
e
im
p
o
r
tan
t to
class
if
y
th
e
n
eg
ativ
e
th
en
to
class
if
y
th
e
p
o
s
itiv
e.
So
,
it is
also
ca
lled
T
NR
.
=
+
(
1
5
)
−
MCC
:
i
t
is
a
m
ea
s
u
r
e
f
o
r
b
in
ar
y
class
if
icatio
n
’
s
q
u
ality
.
I
t
g
iv
es
b
est
r
esu
lt
f
o
r
an
u
n
b
ala
n
ce
d
class
wh
ile
tak
en
in
to
c
o
n
s
id
er
atio
n
T
Ps
,
T
Ns,
FP
s
,
F
Ns.
=
−
√
(
+
)
(
+
)
(
+
)
(
+
)
(
1
6
)
−
Fals
e
p
o
s
itiv
e
r
ate:
i
t
is
t
h
e
p
r
o
b
ab
ilit
y
th
at
p
o
s
itiv
e
r
esu
lt
is
p
r
ed
icted
wh
en
th
e
tr
u
e
v
a
lu
e
is
n
e
g
ativ
e
wh
ich
a
f
alse p
r
ed
ictio
n
.
=
+
(
1
7
)
−
F1
s
co
r
e:
i
t
co
m
b
in
es
th
e
p
r
ec
is
io
n
an
d
r
ec
all
o
f
th
e
s
am
p
les
wh
ich
is
g
iv
en
b
y
th
e
h
ar
m
o
n
ic
m
ea
n
o
f
p
r
ec
is
io
n
an
d
r
ec
all
an
d
is
k
n
o
wn
as
d
ice
s
im
ilar
ity
co
e
f
f
icien
t
(
DSC
)
.
I
t
g
iv
es
b
etter
p
er
f
o
r
m
an
ce
f
o
r
u
n
b
alan
ce
d
d
ataset.
1
=
2
2
+
+
(
1
8
)
3
.
2
.
I
m
ple
m
ent
a
t
io
n r
esu
lt
s
o
f
t
he
1
D
f
e
a
t
ure
-
ba
s
ed
m
a
c
hin
e
lea
rning
a
pp
ro
a
ches
T
h
e
cu
r
r
en
t
s
ec
tio
n
s
h
o
ws
t
h
e
im
p
lem
en
tatio
n
o
f
a
1
D
im
ag
e
-
b
ased
ML
ap
p
r
o
ac
h
f
o
r
p
er
f
o
r
m
in
g
b
in
ar
y
an
d
m
u
lticlas
s
p
r
ed
icti
o
n
.
T
h
er
e
a
r
e
two
s
u
b
s
ec
tio
n
s
–
b
in
a
r
y
p
r
ed
ictio
n
an
d
m
u
lticlas
s
p
r
ed
ictio
n
.
Fo
r
th
e
b
in
ar
y
p
r
e
d
ictio
n
,
th
er
e
ar
e
5
5
0
s
am
p
les
in
ea
ch
ca
teg
o
r
y
as
s
h
o
wn
in
T
ab
le
3
an
d
m
ea
n
class
if
icatio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
40
,
No
.
2
,
No
v
em
b
er
20
25
:
6
5
4
-
6
6
6
660
s
co
r
e
is
s
h
o
wn
in
T
ab
le
4
.
W
h
er
ea
s
,
th
e
m
u
lticlas
s
p
r
ed
ictio
n
u
tili
ze
s
f
o
u
r
ca
teg
o
r
ies wh
ich
h
av
e
5
5
0
s
am
p
le
s
in
ea
ch
ca
teg
o
r
y
in
T
ab
le
5
.
3
.
2
.
1
.
B
ina
ry
predict
io
n
Fo
r
ea
ch
o
f
th
e
two
class
es,
th
e
tr
ain
in
g
a
n
d
test
in
g
s
am
p
les
ar
e
s
p
lit
in
th
e
r
atio
o
f
8
0
:2
0
%.
I
t
r
esu
lts
in
th
e
f
o
r
m
atio
n
o
f
4
4
0
tr
ain
in
g
an
d
1
1
0
test
s
am
p
les
r
esp
ec
tiv
ely
.
T
h
e
tr
ai
n
in
g
s
am
p
les
ar
e
m
ad
e
t
o
u
n
d
er
g
o
tr
ai
n
in
g
with
th
r
ee
ML
tech
n
iq
u
es
in
d
iv
id
u
ally
b
y
u
s
in
g
t
h
e
p
a
r
am
eter
s
m
en
ti
o
n
ed
i
n
s
ec
tio
n
2
.
2
.
3
.
I
t
m
ay
b
e
k
in
d
ly
n
o
ted
th
at
t
h
e
tr
ain
in
g
s
am
p
les
ar
e
in
itial
ly
co
n
v
e
r
ted
to
MFC
C
an
d
p
i
tch
f
ea
tu
r
e
v
ec
t
o
r
s
an
d
th
e
n
f
e
d
to
th
e
ML
alg
o
r
i
th
m
s
.
Du
e
to
th
e
u
s
e
o
f
win
d
o
w
f
o
r
f
ea
tu
r
e
e
x
tr
ac
tio
n
as
m
en
tio
n
ed
in
s
ec
tio
n
2
.
2
,
a
to
tal
o
f
8
,
9
4
1
s
p
ee
ch
f
r
a
m
es a
r
e
g
en
er
ated
f
r
o
m
th
e
1
1
0
h
ea
lth
y
test
s
am
p
les.
Simi
lar
ly
,
a
to
tal
o
f
7
,
6
3
7
f
r
am
es
ar
e
g
en
er
ated
f
r
o
m
1
1
0
p
ath
o
lo
g
ical
test
s
am
p
les.
I
t
is
th
ese
r
e
s
u
ltin
g
test
f
r
am
es
th
at
will
u
n
d
er
g
o
ex
h
au
s
tiv
e
test
in
g
.
T
h
e
test
f
r
am
es
m
en
tio
n
e
d
ab
o
v
e
a
r
e
t
ested
ag
ain
s
t
th
e
K
NN,
Naïv
e
B
ay
es
,
an
d
L
DA
class
if
ier
s
.
T
h
e
p
er
-
class
p
er
f
o
r
m
an
ce
is
h
ig
h
lig
h
ted
in
F
ig
u
r
e
s
3
to
5
.
I
t
is
o
b
s
er
v
e
d
th
at
all
th
e
th
r
ee
class
if
ier
s
p
r
o
v
id
e
s
ig
n
if
ican
tly
lo
w
p
er
-
class
p
er
f
o
r
m
a
n
ce
,
with
6
2
.
4
%
as
s
h
o
wn
in
Fig
u
r
e
3
b
ein
g
th
e
h
ig
h
est f
o
r
t
h
e
h
ea
lth
y
class
an
d
5
0
.
6
%
s
h
o
wn
in
Fig
u
r
e
4
f
o
r
th
e
p
at
h
o
lo
g
ical
class
.
T
h
e
m
ea
n
s
co
r
es d
er
iv
e
d
f
r
o
m
th
ese
p
er
-
class
s
co
r
es
a
r
e
also
h
ig
h
lig
h
ted
in
T
ab
le
4
.
I
t
is
o
b
s
er
v
ed
th
at
K
NN
p
r
o
v
id
es
th
e
h
ig
h
est
ac
cu
r
ac
y
5
7
.
8
9
% in
co
m
p
ar
is
o
n
to
th
e
o
th
er
two
class
if
ier
s
.
T
ab
le
3
.
Nu
m
b
er
o
f
tr
ain
in
g
a
n
d
test
s
am
p
les f
o
r
b
in
a
r
y
p
r
e
d
ictio
n
C
l
a
s
s
Tr
a
i
n
i
n
g
Te
st
i
n
g
To
t
a
l
Te
st
f
r
a
mes
H
e
a
l
t
h
y
4
4
0
1
1
0
5
5
0
8
,
9
4
1
P
a
t
h
o
l
o
g
i
c
a
l
4
4
0
1
1
0
5
5
0
7
,
6
3
7
T
ab
le
4
.
Me
an
class
if
icatio
n
s
co
r
e
o
f
1
D
f
ea
tu
r
e
-
b
ased
b
in
ar
y
p
r
ed
ictio
n
A
l
g
o
r
i
t
h
m
S
e
n
.
A
c
c
.
K
a
p
p
a
Er
r
P
r
e
.
S
p
e
.
M
C
C
FPR
F1
K
N
N
6
7
.
8
9
5
7
.
8
9
15
4
2
.
1
1
5
8
.
2
7
47
1
5
.
2
3
53
6
2
.
7
1
N
a
ï
v
e
B
a
y
e
s
6
2
.
3
9
5
5
.
0
2
9
.
4
3
4
4
.
9
8
5
6
.
2
47
9
.
5
53
5
9
.
1
3
LD
A
5
9
.
6
3
5
4
.
5
5
8
.
6
6
4
5
.
4
5
5
6
.
0
3
49
8
.
6
8
51
5
7
.
7
8
Fig
u
r
e
3
.
Per
-
class
an
d
p
er
-
f
r
a
m
e
class
if
icatio
n
F
i
g
u
r
e
4
.
P
e
r
-
c
la
s
s
a
n
d
p
e
r
-
f
r
a
m
e
c
l
a
s
s
i
f
i
c
a
t
i
o
n
r
es
u
l
t
f
o
r
t
h
e
N
a
ï
v
e
B
a
y
es
m
e
t
h
o
d
r
e
s
u
lt
f
o
r
t
h
e
K
N
N
m
et
h
o
d
Fig
u
r
e
5
.
Per
-
class
an
d
p
er
-
f
r
a
m
e
class
if
icatio
n
r
esu
lt f
o
r
th
e
L
DA
m
eth
o
d
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:
2502
-
4
7
5
2
C
la
s
s
if
ica
tio
n
o
f v
o
ice
p
a
th
o
l
o
g
ies u
s
in
g
o
n
e
d
imen
s
io
n
a
l f
ea
tu
r
e
ve
cto
r
a
n
d
…
(
R
a
n
ita
K
h
u
mu
kc
h
a
m
)
661
T
h
er
ef
o
r
e,
it
is
n
o
te
d
th
at
u
s
e
o
f
1
D
f
ea
tu
r
e
with
ML
class
if
ier
s
ca
n
n
o
t
g
iv
e
g
o
o
d
r
esu
lts
i
n
ca
s
es
o
f
lo
w
n
u
m
b
er
o
f
tr
ain
in
g
s
am
p
l
es.
I
t
is
al
s
o
d
esire
d
to
s
tu
d
y
th
e
p
er
f
o
r
m
an
ce
o
f
2
D
im
ag
e
-
b
ased
d
atasets
with
DL
alg
o
r
ith
m
s
.
T
h
e
y
ar
e
p
er
f
o
r
m
ed
in
s
ec
tio
n
3
.
3
.
3
.
2
.
2
.
M
ulticla
s
s
pre
dict
io
n
Fo
r
th
e
cu
r
r
en
t
m
u
lticlas
s
p
r
ed
ictio
n
also
,
th
e
tr
ain
in
g
an
d
test
s
am
p
les
ar
e
ag
ain
s
p
lit
in
th
e
r
atio
8
0
:2
0
%.
I
t
h
as
also
b
ee
n
d
is
cu
s
s
ed
in
T
a
b
le
2
th
at
t
h
er
e
ar
e
3
8
0
s
am
p
les
i
n
ea
c
h
o
f
t
h
e
f
o
u
r
class
es.
B
y
ap
p
ly
in
g
th
e
a
b
o
v
e
s
p
litt
in
g
r
atio
,
th
e
n
u
m
b
er
o
f
tr
ain
i
n
g
an
d
tes
t
s
am
p
les
in
e
ac
h
cl
ass
ar
e
3
0
4
an
d
7
6
r
esp
ec
tiv
ely
.
Fro
m
all
t
h
e
tr
ain
in
g
s
am
p
les,
f
ea
tu
r
e
v
ec
to
r
s
wh
ich
is
a
co
m
b
i
n
atio
n
o
f
MFC
C
an
d
p
itch
f
ea
tu
r
e
v
ec
t
o
r
s
ar
e
e
x
tr
ac
ted
.
T
h
ese
f
ea
tu
r
e
v
ec
to
r
s
ar
e
u
tili
ze
d
in
tr
ain
i
n
g
th
r
ee
ML
clas
s
if
ier
s
in
d
iv
id
u
ally
by
u
s
in
g
th
e
p
a
r
am
eter
s
m
en
ti
o
n
ed
in
s
ec
tio
n
2
.
2
.
3
.
Als
o
,
d
u
e
to
th
e
u
s
e
o
f
win
d
o
w
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
as
m
en
tio
n
ed
i
n
s
ec
tio
n
2
.
2
,
th
e
n
u
m
b
er
o
f
test
f
r
a
m
es
f
o
r
ea
c
h
o
f
t
h
e
f
o
u
r
class
es
ar
e
5
,
1
9
8
;
5
,
5
1
4
;
3
,
3
9
3
;
a
n
d
5
,
0
9
9
(
s
ee
T
ab
le
5
)
.
T
h
ese
test
f
r
am
es w
ill u
n
d
er
g
o
e
x
h
au
s
tiv
e
test
in
g
.
T
h
e
test
f
r
am
es m
en
tio
n
ed
in
T
ab
le
5
ar
e
test
ed
ag
ain
s
t th
e
KNN,
Naïv
e
B
ay
es,
an
d
L
DA
class
if
ier
s
.
T
h
e
p
er
-
class
p
er
f
o
r
m
a
n
ce
is
h
ig
h
lig
h
ted
in
Fig
u
r
es
6
t
o
8
.
I
t
is
o
b
s
er
v
ed
th
at
th
e
KNN
p
r
o
v
id
ed
t
h
e
b
est
p
er
-
class
ac
cu
r
ac
i
es
f
o
r
h
ea
lth
y
(
i
.
e.
,
4
0
.
2
%),
h
y
p
er
k
in
etic
d
y
s
p
h
o
n
ia
(
i.e
.
,
3
2
.
5
%),
h
y
p
o
k
in
et
ic
d
y
s
p
h
o
n
ia
(
i.e
.
,
5
6
.
5
%).
T
h
e
Naïv
e
B
ay
es
cl
ass
if
ier
p
r
o
v
id
ed
th
e
b
est
p
e
r
f
o
r
m
a
n
ce
f
o
r
th
e
r
ef
lu
x
lar
y
n
g
itis
ca
teg
o
r
y
b
y
d
em
o
n
s
tr
atin
g
a
n
ac
cu
r
ac
y
o
f
4
2
.
9
%.
T
ab
le
5
.
Nu
m
b
er
o
f
tr
ain
in
g
a
n
d
test
s
am
p
les f
o
r
m
u
lticlas
s
p
r
ed
ictio
n
C
l
a
s
s
Tr
a
i
n
i
n
g
Te
st
i
n
g
To
t
a
l
Te
st
f
r
a
mes
H
e
a
l
t
h
y
3
0
4
76
3
8
0
5
,
1
9
8
H
y
p
e
r
k
i
n
e
t
i
c
d
y
sp
h
o
n
i
a
3
0
4
76
3
8
0
5
,
5
1
4
H
y
p
o
k
i
n
e
t
i
c
d
y
s
p
h
o
n
i
a
3
0
4
76
3
8
0
3
,
3
9
3
R
e
f
l
u
x
L
a
r
y
n
g
i
t
i
s
3
0
4
76
3
8
0
5
,
0
9
9
Fig
u
r
e
6
.
Per
-
class
an
d
p
er
-
f
r
a
m
e
class
if
icatio
n
r
esu
lt
f
o
r
th
e
KNN
m
eth
o
d
Fig
u
r
e
7
.
Per
-
class
an
d
p
er
-
f
r
a
m
e
class
if
icatio
n
r
esu
lt f
o
r
th
e
Naïv
e
B
ay
es m
eth
o
d
Fig
u
r
e
8
.
Per
-
class
an
d
p
er
-
f
r
a
m
e
class
if
icatio
n
r
esu
lt f
o
r
th
e
L
DA
m
eth
o
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
40
,
No
.
2
,
No
v
em
b
er
20
25
:
6
5
4
-
6
6
6
662
T
h
e
m
ea
n
s
co
r
es
d
er
iv
e
d
f
r
o
m
th
ese
p
er
-
class
s
co
r
es
ar
e
also
h
ig
h
lig
h
ted
in
T
a
b
le
6
.
I
t
is
o
b
s
er
v
ed
th
at
s
im
ilar
to
th
e
b
in
ar
y
p
r
e
d
ictio
n
ap
p
r
o
ac
h
d
is
cu
s
s
ed
ab
o
v
e,
th
e
K
NN
class
if
ier
d
e
m
o
n
s
tr
ated
th
e
b
est
p
er
f
o
r
m
an
ce
f
o
r
th
e
m
u
lticlas
s
p
r
ed
ictio
n
also
.
I
t
is
also
d
es
ir
ed
to
s
tu
d
y
th
e
p
e
r
f
o
r
m
an
ce
o
f
2
D
im
ag
e
-
b
ased
d
atasets
with
DL
alg
o
r
ith
m
s
.
T
h
ey
ar
e
p
er
f
o
r
m
ed
i
n
s
ec
tio
n
3
.
3
.
T
ab
le
6
.
Me
an
class
if
icatio
n
s
co
r
e
o
f
1
D
f
ea
tu
r
e
-
b
ased
b
in
ar
y
p
r
ed
ictio
n
A
l
g
o
r
i
t
h
m
S
e
n
.
A
c
c
.
K
a
p
p
a
Er
r
o
r
P
r
e
.
S
p
e
.
M
C
C
FPR
F1
k
N
N
3
7
.
8
2
3
7
.
0
2
4
0
.
4
5
6
2
.
9
8
37
7
9
.
0
9
1
6
.
4
7
2
0
.
9
1
3
6
.
1
8
N
a
ï
v
e
B
a
y
e
s
2
8
.
1
4
2
7
.
6
8
4
4
.
7
1
7
2
.
3
2
2
5
.
0
2
7
5
.
9
7
1
4
.
0
3
2
4
.
0
3
2
6
.
1
6
LD
A
3
2
.
7
8
3
2
.
1
8
4
4
.
7
1
6
7
.
8
2
3
0
.
2
3
7
7
.
5
1
1
9
.
9
5
2
2
.
4
9
3
0
.
7
4
3
.
3
.
I
m
ple
m
ent
a
t
io
n r
esu
lt
s
o
f
t
he
2
D
i
m
a
g
e
-
ba
s
ed
deep
lea
rning
a
pp
ro
a
ch
T
h
e
cu
r
r
en
t
s
ec
tio
n
s
h
o
ws
th
e
im
p
lem
en
tatio
n
o
f
a
2
D
im
ag
e
-
b
ased
DL
ap
p
r
o
ac
h
f
o
r
p
er
f
o
r
m
in
g
b
in
ar
y
an
d
m
u
lticlas
s
p
r
e
d
ictio
n
.
T
h
er
e
ar
e
two
s
u
b
s
ec
tio
n
s
–
b
in
a
r
y
p
r
e
d
ictio
n
a
n
m
u
lticlas
s
p
r
e
d
ictio
n
.
Fo
r
th
e
b
in
ar
y
p
r
ed
ictio
n
,
th
er
e
a
r
e
3
8
0
tr
ain
i
n
g
s
am
p
les
in
ea
ch
ca
teg
o
r
y
as
T
a
b
le
7
.
W
h
e
r
ea
s
,
th
e
m
u
lticlas
s
p
r
ed
ictio
n
u
tili
ze
s
f
o
u
r
ca
teg
o
r
ies wh
ich
h
av
e
3
8
0
s
am
p
les i
n
ea
ch
ca
teg
o
r
y
in
T
ab
le
8
.
T
ab
le
7
.
Nu
m
b
er
o
f
tr
ain
in
g
a
n
d
test
s
am
p
les f
o
r
b
in
a
r
y
p
r
e
d
ictio
n
C
l
a
s
s
Tr
a
i
n
i
n
g
Te
st
i
n
g
H
e
a
l
t
h
y
3
0
4
76
P
a
t
h
o
l
o
g
i
c
a
l
3
0
4
76
T
ab
le
8
.
Me
an
class
if
icatio
n
s
co
r
e
o
f
2
D
f
ea
tu
r
e
-
b
ased
b
in
ar
y
p
r
ed
ictio
n
A
l
g
o
r
i
t
h
m
D
a
t
a
s
e
t
S
e
n
.
A
c
c
.
K
a
p
p
a
Er
r
o
r
P
r
e
.
S
p
e
.
M
C
C
FPR
F1
G
o
o
g
L
e
N
e
t
M
.
S
.
9
6
.
0
5
9
6
.
0
5
9
2
.
1
1
3
.
9
5
9
6
.
0
5
9
6
.
0
5
9
2
.
1
1
3
.
9
5
9
6
.
0
5
G
o
o
g
L
e
N
e
t
B
.
S
.
9
6
.
0
5
9
6
.
0
5
9
2
.
1
1
3
.
9
5
9
6
.
0
5
9
6
.
0
5
9
2
.
1
1
3
.
9
5
9
6
.
0
5
G
o
o
g
L
e
N
e
t
A
.
S
.
9
6
.
0
5
9
4
.
7
4
8
9
.
4
7
5
.
2
6
6
.
5
8
9
3
.
4
2
8
9
.
5
9
4
.
8
1
9
4
.
8
1
3
.
3
.
1
.
B
ina
ry
predict
io
n
I
t
ca
n
b
e
s
ee
n
in
T
ab
le
7
th
a
t
th
er
e
ar
e
two
ca
teg
o
r
ies
in
wh
ich
th
er
e
a
r
e
3
8
0
s
am
p
les
ea
ch
.
T
h
e
tr
ain
in
g
an
d
test
in
g
r
atio
wer
e
d
iv
id
ed
in
th
e
r
atio
o
f
8
0
:2
0
%
r
esp
ec
tiv
ely
.
I
t
tr
an
s
lates
to
ar
o
u
n
d
3
0
4
tr
ain
in
g
an
d
7
6
test
s
am
p
les
r
esp
ec
tiv
e
ly
.
T
h
e
tr
ain
i
n
g
an
d
test
s
am
p
les
wer
e
k
ep
t
in
d
if
f
e
r
en
t
f
o
ld
er
s
s
o
th
at
n
o
n
e
o
f
th
e
test
s
am
p
les
wer
e
u
s
ed
(
o
r
s
ee
n
)
d
u
r
in
g
th
e
tr
ai
n
in
g
p
r
o
ce
s
s
.
Fu
r
th
er
m
o
r
e,
th
e
tr
ai
n
in
g
s
am
p
les
wer
e
f
u
r
th
er
d
iv
id
e
d
in
to
tr
ain
in
g
a
n
d
v
alid
atio
n
s
am
p
les
in
th
e
r
atio
8
0
:2
0
%
r
esp
ec
tiv
ely
.
T
h
i
s
m
ea
n
s
th
at
o
u
t
o
f
3
0
4
tr
ain
i
n
g
s
am
p
les th
er
e
a
r
e
2
4
3
ac
t
u
al
tr
ain
in
g
a
n
d
6
1
v
al
id
atio
n
s
am
p
les r
esp
ec
tiv
ely
.
T
h
er
ef
o
r
e,
it
ca
n
b
e
s
u
m
m
a
r
ized
th
at
th
e
r
e
a
r
e
2
4
3
tr
ain
i
n
g
,
6
1
v
alid
atio
n
an
d
7
6
tes
t
s
am
p
les
r
esp
ec
tiv
ely
f
o
r
ea
ch
class
.
T
h
e
p
ar
am
eter
s
m
en
tio
n
ed
in
s
ec
tio
n
3
.
3
(
a
b
o
v
e
)
is
u
s
ed
f
o
r
d
ev
elo
p
in
g
th
e
M.
S
d
atab
ase.
T
h
e
tr
ain
in
g
an
d
v
alid
atio
n
s
am
p
les
ar
e
m
ad
e
to
u
n
d
er
g
o
tr
ain
in
g
b
y
s
ettin
g
th
e
f
o
llo
win
g
p
ar
am
eter
s
:
m
in
ib
atch
s
ize
as
1
6
,
v
alid
atio
n
f
r
eq
u
en
c
y
3
0
an
d
f
lat
lear
n
in
g
r
ate
o
f
0
.
0
0
0
1
.
T
h
e
n
u
m
b
er
o
f
ep
o
ch
s
is
s
et
as
1
5
,
h
o
wev
er
t
h
e
tr
ain
in
g
p
r
o
ce
s
s
is
ter
m
in
ated
wh
en
t
h
e
v
alid
atio
n
ac
cu
r
a
cy
an
d
lo
s
s
cu
r
v
es
b
ec
o
m
e
f
lat.
Fig
u
r
e
9
s
h
o
ws
th
e
tr
ain
in
g
p
r
o
g
r
ess
io
n
f
o
r
th
e
Go
o
g
L
eNe
t
m
o
d
el
with
th
e
M.S
d
ataset
f
o
r
b
in
ar
y
class
p
r
ed
ictio
n
.
Fi
g
u
r
e
1
0
g
iv
es
its
co
n
f
u
s
io
n
m
atr
ix
.
Fo
r
an
ex
ten
s
iv
e
ev
al
u
atio
n
,
th
e
Am
o
r
s
ca
lo
g
r
am
(
A.
S)
an
d
B
u
m
p
s
ca
lo
g
r
am
(
B
.
S)
d
atasets
ar
e
also
d
ev
elo
p
ed
as
s
h
o
wn
in
F
ig
u
r
e
s
11
to
1
4
.
An
o
th
er
two
Go
o
g
L
eNe
t
m
o
d
els
ar
e
also
tr
ain
in
g
with
th
e
s
e
d
atasets
b
y
u
s
in
g
t
h
e
s
am
e
s
e
t
o
f
DL
tr
ain
in
g
p
ar
am
eter
s
.
T
h
e
p
er
class
p
er
f
o
r
m
an
ce
o
f
th
e
Go
o
g
leNe
t
with
th
e
M.
S
d
ataset
is
s
h
o
w
n
b
y
th
e
co
n
f
u
s
io
n
m
atr
ix
o
f
Fig
u
r
e
1
0
.
I
t
is
o
b
s
er
v
ed
th
at
7
3
o
u
t
o
f
7
6
h
ea
lth
y
test
s
am
p
les
ar
e
co
r
r
ec
tly
p
r
ed
icted
,
th
er
e
b
y
g
iv
in
g
a
p
e
r
-
class
ac
cu
r
ac
y
o
f
9
6
.
1
%.
Similar
ly
,
th
e
p
ath
o
lo
g
ical
test
s
am
p
les
ar
e
also
cla
s
s
if
ied
with
a
p
er
-
class
ac
cu
r
ac
y
o
f
9
6
.
1
%.
T
h
e
m
ea
n
class
if
icatio
n
s
co
r
es
ar
e
also
r
ec
o
r
d
ed
in
T
ab
le
8
f
o
r
a
co
m
p
ar
is
o
n
with
o
th
e
r
s
ca
lo
g
r
am
s
s
u
ch
as
th
e
B
.
S
an
d
A.
S
d
ata
s
et.
T
h
e
s
ep
ar
ate
ev
alu
atio
n
o
f
th
e
Go
o
g
L
eNe
t
with
th
e
M.
S.
an
d
B
.
S
d
atasets
s
h
o
ws
a
s
im
ilar
p
er
f
o
r
m
an
ce
,
i.e
.
,
9
6
.
0
5
%
ea
ch
.
T
h
e
v
alu
e
o
f
MCC
an
d
Kap
p
a
ar
e
s
lig
h
tly
lo
w
(
i.e
.
,
9
2
.
1
1
%
ea
ch
)
.
T
h
e
A.
S.
d
ataset
with
th
e
Go
o
g
L
eNe
t
p
r
o
v
i
d
es th
e
lo
west m
ea
n
ac
cu
r
ac
y
,
w
h
ich
is
9
4
.
7
4
%.
3
.
3
.
2
.
M
ulticla
s
s
pre
dict
io
n
T
h
er
e
ar
e
f
o
u
r
class
es
in
th
is
ty
p
e
o
f
p
r
ed
ictio
n
.
T
h
ey
ar
e
–
(
i)
h
ea
lth
y
,
(
i
)
h
y
p
e
r
k
in
etic
d
y
s
p
h
o
n
ia,
(
iii)
h
y
p
o
k
i
n
etic
d
y
s
p
h
o
n
ia,
a
n
d
(
iv
)
r
ef
lu
x
lar
y
n
g
itis
.
T
h
e
s
am
e
tr
ain
in
g
,
v
alid
atio
n
an
d
te
s
t
s
am
p
les
s
p
litt
in
g
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:
2502
-
4
7
5
2
C
la
s
s
if
ica
tio
n
o
f v
o
ice
p
a
th
o
l
o
g
ies u
s
in
g
o
n
e
d
imen
s
io
n
a
l f
ea
tu
r
e
ve
cto
r
a
n
d
…
(
R
a
n
ita
K
h
u
mu
kc
h
a
m
)
663
p
atter
n
m
en
tio
n
ed
i
n
s
ec
tio
n
3
.
3
.
1
is
also
ad
o
p
te
d
h
er
e
.
T
h
er
e
ar
e
2
4
3
tr
ain
in
g
,
6
1
v
ali
d
atio
n
an
d
7
6
test
s
am
p
les
r
esp
ec
tiv
ely
f
o
r
ea
ch
class
.
B
y
ap
p
ly
in
g
th
e
s
am
e
s
et
o
f
p
ar
am
eter
s
m
e
n
tio
n
ed
in
s
ec
tio
n
3
.
3
,
th
e
M.
S,
B
.
S,
an
d
A.
S
d
atasets
ar
e
g
en
er
ated
b
y
u
s
in
g
all
th
e
s
a
m
p
les
m
en
tio
n
ed
in
T
ab
le
9
.
T
h
e
s
am
e
s
et
o
f
DL
tr
ain
in
g
p
a
r
am
eter
s
m
en
tio
n
e
d
in
s
ec
tio
n
3
.
3
.
1
is
u
s
ed
h
e
r
e.
Fig
u
r
e
1
5
s
h
o
ws
th
e
tr
ain
in
g
p
r
o
g
r
ess
o
f
th
e
Go
o
g
leNe
t
with
th
e
M.
S
d
ataset
f
o
r
m
u
lticlas
s
cla
s
s
p
r
ed
ictio
n
.
T
h
e
p
er
class
p
er
f
o
r
m
an
ce
o
f
th
e
Go
o
g
leNe
t
with
th
e
M.
S
d
ataset
is
s
h
o
w
n
b
y
th
e
c
o
n
f
u
s
io
n
m
atr
ix
o
f
Fig
u
r
e
1
6
.
I
t
is
o
b
s
er
v
ed
th
at
s
in
ce
t
h
e
n
u
m
b
er
o
f
class
es
h
as
in
cr
ea
s
ed
in
co
m
p
ar
is
o
n
to
th
e
b
in
ar
y
p
r
ed
ictio
n
,
all
th
e
f
o
u
r
class
es
d
em
o
n
s
tr
ate
ar
o
u
n
d
9
0
%
p
er
-
class
ac
cu
r
ac
y
.
Fo
r
in
s
tan
ce
,
th
e
h
ea
lth
y
test
s
am
p
les
ar
e
class
if
ied
with
an
ac
cu
r
ac
y
o
f
8
8
.
2
%,
th
e
h
y
p
er
k
in
etic
d
y
s
p
h
o
n
ia
an
d
r
e
f
lu
x
lar
y
n
g
itis
ar
e
class
if
ied
with
9
0
.
8
%
ac
cu
r
a
cy
ea
ch
.
Fin
ally
,
th
e
h
y
p
o
k
i
n
etic
d
y
s
p
h
o
n
ia
r
ec
o
r
d
s
a
p
e
r
-
class
ac
cu
r
ac
y
o
f
8
9
.
5
%.
T
h
e
m
ea
n
class
if
icatio
n
s
co
r
es
ar
e
also
r
ec
o
r
d
ed
in
T
ab
le
1
0
f
o
r
a
co
m
p
a
r
is
o
n
with
o
th
er
s
ca
lo
g
r
am
s
s
u
ch
as
th
e
B
.
S
an
d
A.
S
d
ata
s
et.
T
h
e
s
ep
ar
ate
ev
alu
atio
n
o
f
th
e
Go
o
g
L
eNe
t
with
th
e
M.
S.,
B
.
S.
an
d
A.
S.
d
atasets
s
h
o
w
th
at
u
s
e
o
f
M.
S.
with
Go
o
g
L
eNe
t p
r
o
v
id
es th
e
b
est p
er
f
o
r
m
an
ce
o
v
er
9
m
etr
ics.
Fig
u
r
e
1
7
s
h
o
ws
th
e
tr
ain
in
g
p
r
o
g
r
ess
o
f
th
e
Go
o
g
leNe
t
with
th
e
A.
S
d
ataset
f
o
r
m
u
lticlas
s
cla
s
s
p
r
ed
ictio
n
.
T
h
e
p
er
class
p
er
f
o
r
m
an
ce
o
f
th
e
Go
o
g
leNe
t
with
th
e
A.
S
d
ataset
is
s
h
o
wn
b
y
th
e
co
n
f
u
s
io
n
m
atr
ix
o
f
Fig
u
r
e
1
8
.
I
t
is
o
b
s
er
v
ed
th
at
th
e
h
ea
lth
y
test
s
am
p
les
ar
e
clas
s
if
ied
with
an
ac
c
u
r
ac
y
o
f
9
4
.
7
%,
th
e
h
y
p
er
k
in
etic
d
y
s
p
h
o
n
ia
o
f
7
8
.
9
%,
t
h
e
h
y
p
o
k
in
etic
d
y
s
p
h
o
n
ia
o
f
7
7
.
6
%
an
d
r
e
f
lu
x
lar
y
n
g
itis
ar
e
o
f
9
4
.
7
%.
Me
an
wh
ile,
th
e
tr
ain
i
n
g
p
r
o
g
r
ess
o
f
th
e
Go
o
g
leNe
t
with
th
e
B
.
S
d
ataset
f
o
r
f
o
r
m
u
lticlas
s
class
p
r
ed
ictio
n
is
s
h
o
wn
in
Fig
u
r
e
1
9
.
An
d
its
co
n
f
u
s
io
n
m
atr
ix
is
s
h
o
wn
in
Fig
u
r
e
2
0
.
I
t
is
o
b
s
er
v
ed
t
h
at
th
e
h
ea
lth
y
test
s
am
p
les
ar
e
class
if
ied
with
an
ac
cu
r
ac
y
o
f
8
4
.
2
%,
th
e
h
y
p
er
k
in
etic
d
y
s
p
h
o
n
ia
o
f
8
4
.
2
%,
th
e
h
y
p
o
k
in
etic
d
y
s
p
h
o
n
ia
o
f
7
7
.
6
% a
n
d
r
e
f
lu
x
lar
y
n
g
itis
ar
e
o
f
8
9
.
5
%.
Fig
u
r
e
9
.
T
r
ain
in
g
p
r
o
g
r
ess
o
f
th
e
Go
o
g
L
eNe
t w
ith
th
e
p
er
-
class
M.S
d
ataset
f
o
r
b
in
ar
y
cl
ass
p
r
ed
ictio
n
Fig
u
r
e
1
0
.
C
o
n
f
u
s
io
n
m
atr
ix
s
h
o
win
g
th
e
r
esu
lt Go
o
g
L
eNe
t w
ith
th
e
M.
S
Fig
u
r
e
1
1
.
T
r
ain
in
g
p
r
o
g
r
ess
o
f
th
e
Go
o
g
L
eNe
t w
ith
th
e
A.
S
d
ataset
f
o
r
b
in
ar
y
class
p
r
ed
ictio
n
Fig
u
r
e
1
2
.
C
o
n
f
u
s
io
n
m
atr
ix
s
h
o
win
g
th
e
p
er
-
class
r
esu
lt Go
o
g
L
eNe
t w
ith
th
e
A.
S
Evaluation Warning : The document was created with Spire.PDF for Python.