T
E
L
K
O
M
N
I
K
A
T
elec
o
m
m
un
ica
t
io
n,
Co
m
pu
t
ing
,
E
lect
ro
nics
a
nd
Co
ntr
o
l
Vo
l.
18
,
No
.
5
,
Octo
b
er
2
0
2
0
,
p
p
.
2
6
1
2
~
261
7
I
SS
N:
1
6
9
3
-
6
9
3
0
,
ac
cr
ed
ited
First Gr
ad
e
b
y
Kem
en
r
is
tek
d
i
k
ti,
Dec
r
ee
No
: 2
1
/E/KPT
/2
0
1
8
DOI
: 1
0
.
1
2
9
2
8
/TE
L
KOM
NI
K
A.
v
1
8
i5
.
1
3
7
1
7
2612
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//jo
u
r
n
a
l.u
a
d
.
a
c.
id
/in
d
ex
.
p
h
p
/TELK
OM
N
I
K
A
G
ende
r vo
ice clas
sifica
tion wi
th
hu
g
e accura
cy
rate
Mu
sta
fa
S
a
h
ib
S
h
a
r
e
e
f
1
,
Th
u
lfi
q
a
r
Abd
2
,
Ya
q
e
e
n
S
.
Mez
a
a
l
3
1,
2
Al
M
u
th
a
n
n
a
Un
i
v
e
rsity
,
Al
-
M
u
th
a
n
n
a
,
Ira
q
3
M
e
d
ica
l
In
str
u
m
e
n
tatio
n
En
g
i
n
e
e
rin
g
De
p
a
rtme
n
t,
Al
-
Esra
a
Un
iv
e
rsity
Co
ll
e
g
e
,
Ira
q
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
l 2
4
,
2
0
1
9
R
ev
is
ed
Ap
r
6
,
2
0
2
0
Acc
ep
ted
Ap
r
3
0
,
2
0
2
0
G
e
n
d
e
r
v
o
ice
re
c
o
g
n
it
i
o
n
sta
n
d
s
fo
r
a
n
imp
e
ra
ti
v
e
re
se
a
rc
h
field
i
n
a
c
o
u
stics
a
n
d
s
p
e
e
c
h
p
ro
c
e
ss
in
g
a
s
h
u
m
a
n
v
o
ice
s
h
o
ws
v
e
ry
re
m
a
rk
a
b
le
a
sp
e
c
ts.
Th
is
stu
d
y
i
n
v
e
stig
a
tes
s
p
e
e
c
h
sig
n
a
ls
to
d
e
v
ise
a
g
e
n
d
e
r
c
las
sifier
b
y
s
p
e
e
c
h
a
n
a
ly
sis
to
fo
re
c
a
st
th
e
g
e
n
d
e
r
o
f
t
h
e
sp
e
a
k
e
r
b
y
in
v
e
stig
a
ti
n
g
d
iv
e
rse
p
a
ra
m
e
ters
o
f
th
e
v
o
ice
sa
m
p
le.
A
d
a
tab
a
se
h
a
s
2
2
7
0
v
o
ice
s
a
m
p
les
o
f
c
e
leb
rit
ies
,
b
o
t
h
m
a
le
a
n
d
fe
m
a
le.
Th
ro
u
g
h
M
e
l
fre
q
u
e
n
c
y
c
e
p
stru
m
c
o
e
fficie
n
t
(M
F
CC),
v
e
c
to
r
q
u
a
n
ti
z
a
ti
o
n
(VQ
),
a
n
d
m
a
c
h
in
e
lea
rn
in
g
a
lg
o
rit
h
m
(J
4
8
),
a
n
a
c
c
u
ra
c
y
o
f
a
b
o
u
t
1
0
0
%
is
a
c
h
iev
e
d
b
y
t
h
e
p
ro
p
o
se
d
c
las
sifica
ti
o
n
tec
h
n
i
q
u
e
b
a
se
d
o
n
d
a
ta m
in
in
g
a
n
d
Ja
v
a
sc
rip
t
.
K
ey
w
o
r
d
s
:
Au
d
ac
ity
C
las
s
if
icatio
n
ac
cu
r
ac
y
Ma
ch
in
e
lear
n
in
g
alg
o
r
ith
m
(
J
4
8
)
MFC
C
VQ
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
:
Yaq
ee
n
S.
Me
za
al,
Me
d
ical
I
n
s
tr
u
m
en
tatio
n
E
n
g
i
n
ee
r
in
g
Dep
a
r
tm
en
t,
Al
-
E
s
r
aa
Un
iv
er
s
ity
C
o
lleg
e,
B
ag
h
d
ad
,
I
r
aq
.
E
m
ail:
y
ak
ee
n
_
s
b
a
h
@
y
ah
o
o
.
c
o
m
1.
I
NT
RO
D
UCT
I
O
N
I
n
s
p
ee
ch
co
m
m
u
n
icatio
n
,
an
au
d
ito
r
d
o
esn
’
t
m
er
el
y
tr
an
s
late
a
lan
g
u
ag
e
m
ess
ag
e
f
r
o
m
th
e
s
p
ee
ch
s
ig
n
al,
b
u
t
s
im
u
ltan
eo
u
s
ly
h
e/
s
h
e
lik
ewise
d
ed
u
ce
s
p
ar
alin
g
u
is
tic
d
etails
ab
o
u
t
g
en
d
er
,
ag
e
an
d
o
th
e
r
f
ea
tu
r
es
o
f
th
e
talk
er
.
T
h
is
in
f
o
r
m
atio
n
s
o
r
t
s
tan
d
s
f
o
r
s
p
ee
c
h
in
f
o
r
m
atio
n
.
T
h
is
g
en
d
er
d
etec
tio
n
h
as
a
p
p
licatio
n
in
n
u
m
er
o
u
s
f
ield
s
lik
e
g
en
d
er
c
ateg
o
r
izatio
n
o
f
p
h
o
n
e
ca
lls
f
o
r
g
en
d
er
s
en
s
itiv
e
in
v
esti
g
atio
n
s
an
d
class
if
y
in
g
th
e
g
en
d
er
,
a
n
d
elim
in
atin
g
th
e
g
en
d
e
r
p
a
r
ticu
lar
c
o
n
s
titu
en
ts
g
iv
es
g
r
ea
ter
c
o
m
p
r
ess
io
n
r
ate
with
im
p
r
o
v
ed
b
an
d
wid
th
[
1
]
.
B
ased
o
n
ac
o
u
s
tic
f
ea
tu
r
es,
v
o
ice
in
f
o
r
m
atio
n
ca
n
b
e
ex
p
lain
ed
b
y
s
o
m
e
a
co
u
s
tic
f
a
cto
r
s
as
in
th
e
f
u
n
d
a
m
en
tal
f
r
eq
u
en
cy
(
f
0
)
f
o
r
p
er
ce
p
tu
al
r
elev
an
ce
a
n
d
s
p
ec
tr
al
f
o
r
m
a
n
t
f
r
eq
u
en
c
ies
[
2
]
.
Ma
le
h
as
th
e
f
r
eq
u
en
cies le
s
s
th
an
t
h
e
f
em
ale.
Nev
er
th
eless
,
f
o
r
m
an
t
f
r
eq
u
e
n
cy
is
s
o
m
ewh
at
ass
o
ciate
d
with
th
e
v
o
wels
an
d
h
e
n
ce
it
is
tex
t
r
elian
t.
Gen
d
er
class
if
icatio
n
ca
n
b
e
p
r
ep
ar
ed
b
y
p
itch
/f
u
n
d
am
en
tal
f
r
eq
u
en
c
y
ex
tr
ac
te
d
f
r
o
m
d
iv
e
r
s
e
ap
p
r
o
ac
h
es.
Pit
ch
s
tan
d
s
f
o
r
h
ig
h
ly
im
p
er
ativ
e
f
ea
tu
r
e
th
at
ca
n
b
e
g
o
tten
f
r
o
m
d
is
s
im
ilar
m
eth
o
d
s
in
tim
e
an
d
f
r
eq
u
en
cy
d
o
m
ain
s
o
r
b
y
th
e
co
m
b
in
atio
n
o
f
b
o
t
h
o
f
t
h
em
.
T
im
e
d
o
m
ain
ap
p
r
o
ac
h
es
h
a
v
e
all
th
ese
m
eth
o
d
s
th
at
ca
n
b
e
ad
o
p
ted
d
ir
ec
tly
o
n
t
h
e
s
p
ee
ch
s
am
p
les.
T
h
e
s
p
ee
ch
wav
ef
o
r
m
i
s
s
tr
aig
h
tf
o
r
war
d
ly
in
v
esti
g
ated
b
ased
o
n
m
eth
o
d
s
o
f
s
h
o
r
t
tim
e
au
to
c
o
r
r
elat
io
n
,
m
o
d
if
ied
au
to
c
o
r
r
elatio
n
th
r
o
u
g
h
clip
p
in
g
tech
n
iq
u
e,
n
o
r
m
alize
d
cr
o
s
s
co
r
r
elatio
n
f
u
n
ctio
n
,
av
er
a
g
e
m
ag
n
itu
d
e
d
if
f
er
e
n
ce
f
u
n
ctio
n
,
s
q
u
ar
e
d
i
f
f
er
en
ce
f
u
n
ctio
n
etc.
L
ik
ewise,
in
f
r
eq
u
en
cy
d
o
m
ain
a
p
p
r
o
ac
h
es f
o
r
th
e
s
ig
n
al,
th
e
f
r
e
q
u
en
c
y
co
n
te
n
t c
an
b
e
p
r
im
ar
il
y
p
r
o
ce
s
s
ed
an
d
th
en
t
h
e
in
f
o
r
m
atio
n
ca
n
b
e
ex
tr
ac
ted
f
r
o
m
th
e
s
p
ec
tr
u
m
.
T
h
ese
a
p
p
r
o
ac
h
es
co
m
p
r
is
e
h
ar
m
o
n
ic
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
Gen
d
er v
o
ice
cla
s
s
ifica
tio
n
w
ith
h
u
g
e
a
cc
u
r
a
cy
r
a
te
(
Mu
s
ta
f
a
S
a
h
i
b
S
h
a
r
ee
f
)
2613
p
r
o
d
u
ct
s
p
ec
tr
u
m
in
v
esti
g
atio
n
an
d
h
ar
m
o
n
ic
to
s
u
b
h
ar
m
o
n
ic
r
atio
.
T
h
er
e
a
r
e
as
well
s
ev
er
al
m
eth
o
d
s
th
at
d
o
n
’
t c
o
m
e
u
n
d
er
eith
e
r
tim
e
d
o
m
ain
o
r
f
r
e
q
u
en
c
y
d
o
m
ain
a
s
in
wav
elets [
3
,
4
].
2.
L
I
T
E
R
AT
U
RE
R
E
VI
E
W
G
e
n
d
e
r
b
a
s
e
d
r
e
co
g
n
i
t
i
on,
s
p
ee
ch
c
l
a
s
s
i
f
i
c
a
t
i
on
a
nd
p
r
oc
e
s
s
i
ng
wer
e
ad
o
p
ted
f
or
l
ong
t
i
m
e
ag
o
.
To
ap
p
ly
g
en
d
er
r
eco
g
n
i
ti
o
n
,
m
an
y
co
n
ce
p
t
io
n
s
h
av
e
ac
q
u
ir
ed
o
v
er
ti
m
e.
C
u
r
r
en
t
p
a
p
e
r
s
a
b
o
u
t
ge
n
d
e
r
d
e
t
ec
t
i
o
n
h
av
e
s
h
o
wn
t
h
at
v
o
i
ce
ca
n
b
e
alt
e
r
e
d
i
n
t
o
d
i
v
e
r
s
e
p
a
r
a
m
e
t
e
r
s.
F
o
r
em
o
s
t
p
a
r
a
m
e
t
e
r
s
ar
e
p
i
t
ch
an
d
fr
e
que
n
c
y
.
C
l
a
s
s
i
f
i
c
a
ti
on
ca
n
b
e
o
r
g
an
i
ze
d
f
o
r
d
i
s
t
i
n
gu
i
s
h
i
n
g
f
e
m
a
l
e,
m
a
l
e
an
d
c
h
i
l
d
r
e
n.
Firstl
y
,
th
e
r
e
co
g
n
i
t
i
o
n
s
y
s
tem
w
a
s
p
r
e
p
a
r
e
d
w
it
h
t
h
e
t
r
a
i
n
i
ng
d
a
t
a.
T
h
en
,
t
e
s
tin
g
d
a
t
a
h
a
v
e
p
r
esen
ted
a
n
d
as
s
e
ss
e
d
b
ased
o
n
s
y
s
t
em
p
e
r
f
o
r
m
an
ce
f
o
r
t
h
e
se
d
a
t
a.
T
he
ac
q
u
i
r
ed
c
o
n
s
e
q
u
en
ce
s
h
av
e
b
ee
n
d
i
s
s
i
m
i
l
ar
f
or
d
i
ve
r
s
e
p
r
o
c
e
d
u
r
e
s
.
T
h
ey
h
av
e
g
en
er
ated
d
i
v
e
r
s
e
p
e
r
ce
p
t
i
b
l
e
c
o
n
s
e
q
u
en
ce
s
a
t
d
i
ve
r
s
e
pe
r
i
o
d
s
.
G
en
d
er
b
a
s
e
d
c
l
a
s
s
i
f
i
c
a
t
i
o
n
b
ased
o
n
f
u
n
d
am
e
n
tal
f
r
e
q
u
e
n
c
y
[
5
]
a
n
d
p
i
t
ch
wi
t
h
n
u
m
e
r
o
u
s
t
r
a
i
n
i
n
g
an
d
t
e
s
tin
g
d
a
t
a
h
as
s
h
o
wn
t
h
a
t
L
o
g
i
s
t
i
c
r
e
g
r
e
s
s
i
on
h
as
b
e
en
t
h
e
fi
n
est
a
l
g
o
r
it
h
m
with
a
cc
u
r
ac
y
o
f
9
2
%
th
a
n
r
an
d
o
m
f
o
r
e
s
t
.
A
da
B
oo
s
t
f
or
v
o
i
ce
d
a
t
a
wit
h
i
d
e
n
tical
l
a
n
g
ua
g
e
p
e
r
f
o
r
m
s
b
e
t
t
er
i
n
t
h
e
c
a
s
e
o
f
R
an
d
o
m
Fo
r
e
s
t
a
l
g
o
r
it
h
m
with
a
c
cu
r
acy
of
9
3
%
.
Fo
r
s
p
ee
ch
r
eco
g
n
i
ti
on,
r
a
n
d
o
m
f
o
r
e
st
o
u
t
f
its
ag
r
ee
ab
ly
in
ac
co
r
d
an
ce
with
f
u
n
d
am
e
n
tal
f
r
e
q
u
e
n
c
y
an
d
p
it
c
h
f
o
r
c
a
t
e
g
o
r
izin
g
f
e
m
a
l
e,
m
a
l
e
an
d
a
ch
i
l
d
[
6
]
.
A
dd
e
d
t
u
n
i
ng
s
tan
d
s
f
o
r
th
e
b
i
nn
i
ng
m
et
h
o
d
f
o
r
d
e
v
e
l
o
p
in
g
t
h
e
e
f
f
i
c
i
e
n
cy
with
d
e
s
i
r
e
d
f
a
ll
o
u
ts
.
V
o
i
ce
b
a
s
e
d
w
o
r
d
e
x
tr
a
c
t
i
o
n
l
ab
v
i
ew
[
7
]
o
p
er
ates
well
f
or
v
o
ic
e
c
l
a
s
s
i
f
i
ca
t
i
o
n
.
I
t
h
as
b
e
t
t
er
r
e
s
u
l
t
s
f
o
r
v
o
w
e
l
s
ex
t
r
ac
tin
g
i
n
m
a
l
e
test
er
s
.
W
h
e
n
t
he
t
e
s
t
e
rs
h
a
v
e
b
e
e
n
tr
ain
ed
a
nd
i
n
v
esti
g
ate
d
,
i
t
c
a
p
a
b
l
y
g
i
v
es
a
m
ea
n
i
n
g
f
u
l
o
u
tc
o
me
.
I
t
h
as
a
s
well
p
er
c
e
i
v
ed
t
h
a
t
b
y
a
gg
r
eg
a
t
in
g
an
un
v
o
i
ced
fr
a
g
m
e
n
t
i
n
sp
e
e
c
h
r
e
l
at
ed
t
o
a
s
o
u
n
d
o
f
‘
s’
v
a
l
ue
o
f
p
it
c
h
u
p
s
u
r
g
e
s
ob
s
tr
uc
t
in
g
a
g
en
d
e
r
d
e
t
e
c
t
i
ng
f
o
r
m
a
le
s
a
m
p
l
es.
Co
r
re
s
po
n
d
i
n
g
l
y
,
by
i
nc
r
e
a
s
ed
v
o
i
ce
fr
a
g
m
en
t
of
s
p
ee
c
h
as
in
’
a
’
,
it
dr
o
p
s
t
he
p
i
t
ch
v
a
l
u
e
an
d
it
is
u
n
s
u
cc
es
s
f
u
l
f
o
r
c
l
as
s
if
y
in
g
i
t
as
t
he
sp
eak
er
talk
s
d
u
al
d
i
s
s
i
m
i
l
ar
t
on
e
s
.
Sp
ee
ch
r
e
c
o
g
n
i
z
i
n
g
s
y
s
t
e
m
s
i
n
a
d
u
l
t
h
as
i
m
p
u
l
s
i
v
e
an
d
v
o
ca
l
l
en
g
t
h
c
h
a
n
g
e
s
.
T
h
ey
a
r
e
ab
le
to
s
o
u
n
d
as
i
n
m
a
l
e
an
d
f
e
m
a
l
e
.
C
o
n
s
eq
u
en
tly
,
i
t
h
as
b
e
e
n
h
a
r
d
t
o
c
a
t
e
g
o
r
ize
t
he
b
o
t
h
g
e
n
d
e
r
s
.
Sev
er
al
f
e
m
i
n
in
e
v
o
i
ces
h
av
e
b
ee
n
d
i
f
f
i
c
u
l
t
f
o
r
i
n
v
e
st
ig
atin
g
ba
s
ed
on
t
he
p
it
ch
[
8
]
as
in
in
v
e
s
ti
g
a
tio
n
of
s
in
g
le
f
ac
e
t
of
w
o
m
a
n
ly
v
o
i
ce
s
th
at
d
o
esn
’
t
s
a
t
i
s
f
y
th
e
r
eq
u
i
r
e
m
en
t
s
.
B
a
s
e
d
on
[
8
]
,
th
e
f
e
m
i
n
in
e
v
o
i
ce
s
h
o
u
l
d
b
e
r
ec
o
g
n
i
z
ed
wi
t
h
d
i
s
s
i
m
i
la
r
p
a
r
a
m
e
t
e
r
s
as
co
m
p
ar
ed
wi
th
m
a
l
e
s
u
ch
as
s
h
r
i
ll
,
h
i
g
h
-
p
it
ch
,
em
o
tiv
e
an
d
s
w
o
op
y
p
a
r
a
m
e
t
e
r
s
.
T
he
s
e
a
r
e
d
i
s
s
i
m
i
lar
p
a
r
a
m
e
t
e
r
s
f
o
r
f
e
m
i
n
in
e
in
d
iv
id
u
als
,
an
d
t
h
e
y
d
i
ff
er
f
r
o
m
f
em
ale
t
o
a
n
o
t
h
e
r
.
T
h
er
ef
o
r
e,
d
a
t
a
se
t
m
u
s
t
be
s
o
r
t
ed
o
ut
ba
s
ed
on
t
h
i
s
ad
o
p
ted
c
l
a
s
s
i
f
i
c
a
t
i
on
o
f
m
a
s
cu
lin
e
an
d
f
e
m
i
n
in
e.
In
[
9
]
,
fu
n
da
m
en
t
a
l
f
r
eq
u
en
cy
h
as
a
ve
r
b
a
l
gr
o
u
p
in
g
with
n
o
n
l
i
n
g
u
i
s
t
i
c
an
d
p
a
r
a
l
i
n
g
u
i
s
t
i
c
d
a
t
a
ab
o
u
t
s
p
ea
k
er
.
T
h
e
se
t
h
r
ee
i
s
s
u
e
s
r
e
l
a
t
e
t
o
m
a
s
cu
lin
e
an
d
f
e
m
i
n
in
e
,
a
nd
i
t
a
s
well
r
elie
s
on
huge
p
i
t
ch
a
n
d
t
one
o
f
t
he
t
a
lk
e
r
.
I
t
w
a
s
a
ch
i
e
v
e
d
f
o
r
s
ettin
g
a
fr
e
q
u
en
cy
ir
r
es
p
ec
t
iv
e
o
f
an
y
acq
u
ai
n
tan
ce
ab
o
u
t
r
an
g
e
a
n
d
s
y
l
l
a
b
l
e
-
ex
t
e
r
n
al
d
a
ta
.
A
c
c
o
r
d
i
n
g
l
y
,
a
s
p
ea
k
e
r
v
o
i
ce
d
i
f
f
e
r
s
b
e
t
w
een
h
i
g
h
an
d
l
ow
p
i
t
ch
es
am
o
n
g
u
tter
e
r
s
.
Glo
ttal
-
p
u
ls
e
r
ate
(
GPR
)
an
d
v
o
ca
l
-
t
r
ac
t
len
g
th
(
VT
L
)
h
a
v
e
ass
o
ciate
d
with
ag
e,
s
ex
a
n
d
s
ize
o
f
a
s
p
ea
k
er
.
I
t
h
as
b
ee
n
n
o
t
ce
r
tain
ab
o
u
t
th
e
way
s
f
o
r
th
e
d
u
al
f
ac
to
r
s
to
b
e
in
teg
r
ated
f
o
r
af
f
e
ctin
g
th
e
p
er
ce
p
tio
n
o
f
s
p
ea
k
er
s
ex
,
ag
e
an
d
s
ize.
I
n
[
1
0
]
,
ex
p
er
im
en
ts
h
av
e
co
n
d
u
cted
to
ev
alu
ate
th
e
in
ter
f
ac
e
in
f
lu
en
ce
f
o
r
GPR
an
d
VT
L
u
p
o
n
f
in
d
i
n
g
s
o
f
s
p
e
ak
er
s
ex
,
s
ize
an
d
ag
e.
V
o
wels
h
av
e
b
ee
n
s
ca
led
f
o
r
ch
ar
ac
t
er
izin
g
in
d
i
v
id
u
als
with
v
ar
ied
GPR
s
an
d
VT
L
s
,
co
n
tain
in
g
n
u
m
er
o
u
s
am
o
u
n
t
s
o
v
er
th
a
n
th
e
s
tan
d
ar
d
r
an
g
e
o
f
t
h
e
p
o
p
u
lace
.
Sp
ec
tato
r
s
h
av
e
b
ee
n
r
eq
u
est
ed
to
esti
m
ate
a
s
ize
an
d
s
ex
/ag
e
o
f
a
talk
er
.
T
h
e
d
is
cr
im
in
atio
n
s
o
f
s
p
ea
k
er
s
ize
ex
p
lain
th
at
VT
L
h
o
ld
s
a
r
esil
ien
t
im
p
ac
t
o
n
ap
p
ar
en
t
s
p
ea
k
er
s
ize.
T
h
e
f
allo
u
t
s
f
o
r
s
ex
an
d
ag
e
ca
teg
o
r
y
(
wo
m
a
n
,
m
a
n
,
g
ir
l
o
r
b
o
y
)
ex
p
lain
th
at
f
o
r
v
o
wels
with
GPR
an
d
VT
L
m
a
g
n
itu
d
es
in
th
e
s
tan
d
ar
d
r
an
g
e.
Fin
d
in
g
s
o
f
s
p
ea
k
e
r
a
g
e
an
d
s
ex
h
a
v
e
af
f
ec
te
d
eq
u
iv
alen
tly
b
y
GPR
an
d
VT
L
.
Fo
r
v
o
wels
with
ir
r
eg
u
lar
g
r
o
u
p
in
g
s
o
f
s
m
all
G
PR
s
an
d
d
im
in
u
tiv
e
VT
L
s
,
th
e
VT
L
d
ata
s
ee
m
to
h
a
v
e
im
p
a
cts
o
n
d
eter
m
in
in
g
th
e
s
ex
/ag
e
ju
d
g
e
m
en
t.
I
n
[
1
1
]
,
t
h
e
au
to
m
atic
v
o
ice
d
i
s
o
r
d
er
class
if
icatio
n
s
y
s
tem
was p
r
o
jecte
d
b
ased
o
n
f
ir
s
t d
u
a
l f
o
r
m
an
ts
o
f
v
o
wels.
Fiv
e
ca
teg
o
r
ies
o
f
v
o
ice
s
y
n
d
r
o
m
e
o
f
p
a
r
aly
s
is
,
p
o
ly
p
,
s
u
lcu
s
an
d
cy
s
t,
h
a
v
e
b
ee
n
em
p
lo
y
e
d
in
th
e
test
s
.
Vo
iced
Ar
ab
ic
d
ig
it
s
f
r
o
m
th
e
v
o
ice
d
is
o
r
d
er
ly
i
n
d
iv
id
u
als
h
a
v
e
v
er
if
ie
d
f
o
r
a
n
in
p
u
t.
T
h
e
in
itial
f
o
r
m
an
t
an
d
2
nd
f
o
r
m
a
n
t
h
a
v
e
ex
tr
ac
ted
f
r
o
m
th
e
[
Fath
a]
an
d
[
Kasr
a]
v
o
wels t
h
at
h
av
e
b
ee
n
ex
is
tin
g
in
Ar
ab
ic
d
ig
its
.
T
h
e
f
o
u
r
f
ea
tu
r
es
h
av
e
b
ee
n
th
en
em
p
lo
y
ed
f
o
r
ca
teg
o
r
izin
g
th
e
v
o
ice
d
is
o
r
d
er
b
y
m
ea
n
s
o
f
d
u
al
ca
teg
o
r
ies
o
f
class
if
icatio
n
s
y
s
tem
s
:
v
ec
to
r
q
u
an
tizatio
n
(
VQ)
an
d
n
eu
r
al
n
etwo
r
k
s
(
NNs).
I
n
th
e
test
s
,
NN
im
p
lem
en
ts
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
as
co
m
p
ar
ed
with
VQ.
Fo
r
f
em
ale
an
d
m
ale
talk
er
s
,
th
e
class
if
icat
io
n
p
er
ce
n
tag
es
h
a
v
e
b
ee
n
6
7
.
8
6
%
an
d
5
2
.
5
%,
b
y
m
ea
n
s
o
f
N
N.
A
f
in
est
class
if
icatio
n
r
ate
h
as
b
ee
n
7
8
.
7
2
%
f
o
r
f
em
in
in
e
s
u
lcu
s
d
is
o
r
d
e
r
.
G
e
n
d
e
r
i
d
en
ti
f
i
c
a
t
i
on
by
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(S
V
M)
[
12]
s
h
o
w
s
t
h
at
g
en
d
er
s
p
ee
c
h
i
s
ev
a
l
ua
ted
by
n
u
me
r
o
u
s
sp
e
e
c
h
a
p
p
lia
n
c
e
s
a
s
in
c
o
m
p
r
e
s
s
e
d
s
pe
e
ch
,
t
e
l
ep
h
o
n
e
s
p
ea
k
i
ng
a
n
d
v
a
r
i
a
n
ce
i
n
l
an
g
ua
g
es,
e
tc.
I
t
tr
a
n
s
f
er
s
t
h
at
m
a
s
cu
lin
e
vo
i
ce
f
r
om
p
it
ch
,
pe
r
i
od
a
n
d
M
e
l
-
fr
eq
u
e
n
cy
h
as
b
e
en
wi
th
in
10
0
-
1
4
6
H
z
r
an
g
e
an
d
i
n
t
he
f
e
m
a
l
e
w
i
t
h
i
n
188
-
221
H
z
.
A
t
th
is
p
o
in
t,
t
he
v
o
i
c
e
h
a
s
b
e
en
s
e
p
a
r
a
t
ed
ba
s
ed
on
t
h
e
e
x
t
r
ac
t
ed
f
ea
tu
r
e
as we
ll a
s
f
r
eq
u
e
n
cy
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
,
Vo
l.
18
,
No
.
5
,
Octo
b
e
r
2
0
2
0
:
261
2
-
261
7
2614
Gau
s
s
ian
m
ix
tu
r
e
m
o
d
el
(
G
MM
)
ba
s
e
d
g
en
d
er
c
l
a
s
s
i
f
i
c
a
ti
on
[
1
3
]
,
s
u
gge
s
ts
t
h
a
t
sp
e
e
c
h
ca
n
b
e
ex
a
m
in
ed
b
ased
on
a
g
e,
w
o
r
d
s
,
e
t
c.
M
i
x
t
u
r
e
m
o
d
el
h
as
b
e
e
n
a
d
o
p
t
e
d
with
r
e
c
o
g
n
i
z
in
g
ac
c
u
r
acy
up
t
o
9
8%
.
I
t
s
t
an
d
s
f
o
r
t
h
e
e
f
f
e
c
t
u
a
l
t
e
c
h
n
iq
u
e
f
or
s
p
ee
c
h
de
t
e
c
ti
n
g
an
d
g
en
d
e
r
i
n
v
e
s
t
i
g
atio
n
.
C
l
a
s
s
i
f
i
c
a
ti
o
n
h
as
b
ee
n
d
ep
en
d
e
n
t
on
jo
i
n
t
f
a
c
t
o
r
s
of
p
i
t
ch
a
n
d
r
e
l
a
t
i
v
e
s
pe
c
t
r
al
p
e
r
ce
p
t
u
a
l
li
n
ear
p
r
e
d
i
c
ti
v
e
(
R
a
s
t
a
-
Plp
)
fa
c
t
o
r
f
o
r
m
o
d
elin
g
m
a
s
cu
lin
e
an
d
f
e
m
i
n
in
e
s
p
e
ec
h
.
T
h
is
r
esear
ch
wo
r
k
ex
am
in
es
s
p
ee
ch
s
ig
n
als
to
d
ev
elo
p
a
g
en
d
er
class
if
ier
b
y
s
p
ee
ch
a
n
aly
s
is
f
o
r
f
o
r
ec
asti
n
g
th
e
g
en
d
e
r
o
f
th
e
s
p
ea
k
er
b
y
in
v
esti
g
atin
g
d
i
v
er
s
e
p
ar
am
eter
s
o
f
th
e
v
o
ice
s
a
m
p
le.
B
y
m
ea
n
s
o
f
M
F
C
C
,
VQ,
an
d
m
ac
h
in
e
lea
r
n
in
g
alg
o
r
ith
m
(
J
4
8
)
,
an
ac
cu
r
ac
y
o
f
ap
p
r
o
x
im
ately
1
0
0
%
h
as
r
ea
lized
b
y
th
e
p
r
o
jecte
d
class
if
icatio
n
tech
n
iq
u
e
b
ased
o
n
d
ata
m
in
i
n
g
an
d
J
av
a
s
cr
ip
t
a
p
p
lied
o
n
a
d
atab
ase
with
2
2
7
0
v
o
ice
s
am
p
les o
f
ce
leb
r
it
ies
.
3.
M
E
L
F
R
E
QU
E
N
C
Y
C
E
P
S
T
R
UM
CO
E
F
F
I
C
IE
N
T
In
t
h
i
s
s
t
u
d
y
,
MF
CC
p
ro
c
e
d
u
re
h
a
s
b
ee
n
e
m
p
lo
y
ed
f
o
r
e
x
t
r
a
c
tin
g
f
e
at
u
r
e
s
f
r
o
m
a
s
p
e
e
c
h
s
i
g
n
a
l
a
n
d
c
o
m
p
a
r
e
t
h
e
u
n
i
d
e
n
tifie
d
s
p
e
a
k
e
r
w
it
h
t
h
e
c
u
r
r
en
t
s
p
e
a
k
e
r
i
n
a
d
a
ta
b
a
s
e
.
F
i
g
u
r
e
1
c
l
a
r
i
f
i
es
a
f
u
l
l
p
i
p
e
l
i
n
e
o
f
M
FC
C
t
e
c
h
n
i
q
u
e
[
1
4
]
.
MF
CC
t
ec
hn
i
que
h
as
fr
e
q
u
en
tly
em
p
l
o
y
ed
f
o
r
g
e
ne
r
atin
g
a
f
i
n
g
e
r
p
r
i
n
t
o
f
t
h
e
s
o
u
n
d
f
il
e
s
.
MF
CC
h
a
s
b
e
e
n
d
e
p
e
n
d
e
n
t
o
n
r
e
c
o
g
n
ized
d
i
s
s
im
i
l
ar
ity
o
f
hu
ma
n
o
i
d
ea
r
’
s
c
r
i
t
i
c
a
l
b
a
n
d
w
i
d
t
h
f
r
e
qu
e
n
c
i
e
s
w
i
t
h
s
p
a
ce
d
f
ilt
er
s
in
l
in
ea
r
w
ay
a
t
l
o
w
e
r
f
r
e
qu
e
n
c
i
e
s
.
Als
o
,
it
i
s
em
p
l
o
y
ed
l
o
g
a
r
i
t
h
m
i
c
all
y
a
t
h
i
g
h
e
r
f
r
e
qu
e
n
c
i
e
s
f
o
r
c
a
p
t
u
r
in
g
t
h
e
im
p
e
r
a
t
iv
e
f
e
a
t
u
r
e
s
of
s
p
e
ec
h
.
R
e
p
o
r
t
e
d
r
e
s
e
a
r
c
h
es
r
e
v
e
aled
t
h
a
t
hu
ma
n
o
id
p
e
r
ce
p
ti
o
n
o
f
t
h
e
f
r
e
qu
e
n
c
y
s
u
b
j
e
c
t
s
fo
r
s
p
e
e
c
h
s
i
gn
al
s
d
o
esn
’
t
h
a
v
e
th
e
l
i
n
ea
r
s
c
al
e
.
T
h
er
ef
o
r
e,
a
l
l
t
o
n
e
s
h
a
v
e
a
n
a
u
t
h
e
n
tic
f
r
e
qu
e
n
c
y
,
f
,
d
eter
m
i
n
ed
i
n
H
z
,
w
h
i
l
e
a
s
p
e
c
i
f
ie
d
p
it
c
h
h
as
d
ete
r
m
i
n
ed
b
a
s
ed
o
n
a
s
c
al
e
t
e
r
m
e
d
as
M
e
l
s
c
al
e
.
T
h
e
M
e
l
-
f
r
e
qu
e
n
c
y
s
c
al
e
i
s
li
n
ea
r
f
r
e
qu
e
n
c
y
s
p
a
c
i
n
g
a
t
l
o
w
er
th
an
1
0
0
0
H
z
a
n
d
a
l
o
g
a
r
it
h
mi
c
s
p
a
ci
n
g
b
ey
ond
1000
H
z
.
T
h
e
p
it
c
h
o
f
a
1
k
H
z
t
o
n
e
,
40
dB
h
ig
h
e
r
th
an
t
h
e
p
e
r
ce
p
t
u
a
l
h
ea
r
i
n
g
t
h
r
e
s
h
o
l
d
,
is
f
ea
s
ib
ly
s
t
at
ed
a
s
1
000
M
e
l
s
a
s
a
r
e
f
e
r
e
n
c
e
p
o
i
n
t
.
Fig
u
r
e
1
.
Pip
elin
e
o
f
MFC
C
T
h
e
fo
ll
o
w
i
n
g
eq
u
atio
n
is
f
o
r
c
a
l
cu
l
atin
g
t
h
e
M
e
l
s
f
o
r
th
e
s
p
e
ci
f
i
c
f
r
e
qu
e
n
c
y
:
(
)
=
2595
∗
10
(
1
+
/
700
)
.
T
h
e
b
r
i
e
f
a
b
o
u
t
MF
CC
p
r
o
ce
d
u
r
es
h
as
p
r
e
s
e
n
t
e
d
in
F
i
g
u
r
e
2
.
A
s
p
eec
h
w
a
v
e
f
o
r
m
h
as
c
o
l
le
c
ted
f
o
r
r
e
m
o
v
i
n
g
a
c
o
u
s
t
i
c
a
l
i
n
t
e
r
f
e
r
e
n
c
e
o
r
s
il
e
n
c
e
t
h
a
t
is
f
ea
s
ib
ly
e
x
i
s
t
i
ng
i
n
t
h
e
b
e
g
i
n
n
i
n
g
o
r
e
n
d
in
g
o
f
a
s
o
u
n
d
f
il
e
.
A
w
i
n
d
o
w
i
n
g
b
l
o
c
k
re
d
u
c
es
t
h
e
s
i
gn
a
l
d
i
s
c
on
t
i
n
u
i
t
i
e
s
t
h
r
o
u
g
h
ta
p
e
r
i
n
g
t
h
e
s
t
a
r
tin
g
a
nd
t
e
r
m
in
atio
n
o
f
e
a
c
h
f
r
a
m
e
t
o
z
e
r
o
.
T
h
e
F
F
T
b
l
o
c
k
t
r
a
n
s
f
o
rms
a
l
l
f
r
a
m
es
f
r
o
m
t
i
m
e
t
o
f
r
e
qu
e
n
c
y
d
o
ma
i
n
.
A
s
i
g
n
a
l
h
a
s
s
c
h
em
e
d
i
n
t
h
e
M
e
l
-
f
r
e
qu
e
n
c
y
w
r
a
pp
i
n
g
b
l
o
c
k
i
n
c
o
n
tr
ad
ictio
n
o
f
t
h
e
M
e
l
s
p
ec
t
r
um
t
o
m
ir
r
o
r
h
u
ma
n
o
id
h
ea
r
i
n
g
.
I
n
an
e
n
d
i
ng
s
t
e
p,
t
h
e
c
e
ps
t
r
u
m
an
d
th
e
M
e
l
-
s
p
ec
t
r
u
m
s
c
a
l
e
can
a
d
a
p
t
b
a
c
k
to
t
y
p
i
c
a
l
f
r
eq
u
e
n
cy
s
c
al
e
.
T
h
i
s
s
p
ect
r
um
h
a
s
a
n
o
b
l
e
d
e
p
i
c
t
i
o
n
o
f
t
h
e
s
p
ec
t
r
a
l
f
e
a
t
u
r
e
s
of
a
s
i
g
n
a
l
t
h
a
t
h
a
s
b
e
e
n
i
m
p
o
r
tan
t f
o
r
s
i
g
n
i
f
y
i
n
g
a
n
d
i
de
n
ti
fy
i
ng
f
e
a
t
u
r
e
s
o
f
a
s
p
e
a
k
e
r
.
O
n
c
e
t
h
e
f
i
n
g
e
r
p
r
i
n
t
h
as
p
r
o
du
c
ed
,
t
h
e
a
c
o
u
s
ti
c
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
Gen
d
er v
o
ice
cla
s
s
ifica
tio
n
w
ith
h
u
g
e
a
cc
u
r
a
cy
r
a
te
(
Mu
s
ta
f
a
S
a
h
i
b
S
h
a
r
ee
f
)
2615
v
ec
t
o
r
can
b
e
k
e
p
t
a
s
a
r
e
f
e
r
e
n
c
e
i
n
a
d
a
t
a
b
a
s
e
.
Fu
r
th
er
m
o
r
e,
it
s
c
o
n
s
e
qu
en
tial
v
ec
t
o
r
can
b
e
c
o
m
p
a
r
e
d
with
t
h
o
s
e
i
n
a
d
a
t
a
b
a
s
e
.
An
o
t
h
er
t
im
e
b
y
m
ea
n
s
o
f
e
u
c
li
d
i
a
n
d
i
s
ta
n
c
e
t
ec
hn
i
qu
e
a
n
d
an
a
p
p
ro
p
r
iate
m
a
t
c
h
i
n
g
can
b
e
co
n
clu
d
ed
.
T
h
i
s
p
r
o
ce
s
s
h
as
te
r
m
ed
a
s
f
e
at
u
r
e
m
a
t
c
h
i
n
g.
Fig
u
r
e
2
.
MFC
C
d
iag
r
am
4.
V
E
C
T
O
R
Q
U
AN
T
I
Z
A
T
I
ON
T
h
e
s
p
ea
k
er
r
ec
o
g
n
itio
n
s
y
s
tem
wo
u
ld
b
e
ca
p
ab
le
f
o
r
esti
m
atin
g
p
r
o
b
ab
ilit
y
d
is
tr
ib
u
tio
n
s
o
f
th
e
p
r
o
ce
s
s
ed
f
ea
tu
r
e
v
ec
t
o
r
s
.
Sto
r
ag
e
f
o
r
ea
c
h
d
is
tin
ct
v
ec
to
r
th
at
p
r
o
d
u
ce
s
f
r
o
m
th
e
m
o
d
e
tr
ain
in
g
is
in
to
ler
ab
le,
e
v
en
as
t
h
ese
d
is
tr
ib
u
tio
n
s
ar
e
d
ef
in
ite
o
v
e
r
a
b
ig
d
im
en
s
io
n
al
s
p
ac
e.
I
t
h
as
b
ee
n
f
r
eq
u
en
tly
s
im
p
ler
to
b
eg
i
n
th
r
o
u
g
h
q
u
a
n
tizin
g
ev
er
y
f
ea
tu
r
e
v
ec
to
r
to
o
n
e
o
f
a
r
ea
s
o
n
ab
ly
in
s
ig
n
if
ica
n
t
a
m
o
u
n
t
o
f
tem
p
late
v
ec
to
r
s
,
with
v
ec
to
r
q
u
an
tizat
io
n
.
VQ
p
r
o
ce
s
s
es
a
h
u
g
e
s
et
o
f
f
ea
tu
r
e
v
ec
to
r
s
an
d
g
en
er
at
es
a
r
ed
u
ce
d
s
et
o
f
m
ea
s
u
r
e
v
ec
to
r
s
th
at
ch
ar
ac
te
r
izes
th
e
d
is
tr
ib
u
tio
n
ce
n
tr
o
id
s
.
T
h
e
VQ
m
eth
o
d
h
as
a
f
ea
s
ib
ilit
y
f
o
r
e
x
tr
ac
tin
g
a
s
m
all
am
o
u
n
t
o
f
d
escr
ip
tiv
e
f
ea
tu
r
e
v
ec
to
r
s
as
th
e
ef
f
ec
tu
al
to
o
l
f
o
r
d
is
tin
g
u
is
h
in
g
t
h
e
s
p
ea
k
er
s
p
ec
if
ic
f
ea
tu
r
es.
Acc
o
r
d
in
g
ly
,
s
to
r
ag
e
o
f
ea
ch
g
e
n
er
ated
v
ec
to
r
f
r
o
m
th
e
tr
ain
in
g
h
as
b
ee
n
in
t
o
ler
ab
le.
B
ased
o
n
ad
o
p
ted
t
r
ain
in
g
d
ata,
f
ea
t
u
r
e
s
h
av
e
co
llected
f
o
r
g
en
er
ati
n
g
a
c
o
d
eb
o
o
k
f
o
r
ev
er
y
s
p
e
ak
er
.
T
h
e
d
ata,
i
n
r
ec
o
g
n
itio
n
s
tep
f
r
o
m
th
e
i
n
v
esti
g
ated
s
p
ea
k
er
,
will
b
e
co
m
p
ar
ed
to
a
c
o
d
eb
o
o
k
o
f
ea
ch
s
p
ea
k
er
,
an
d
a
d
if
f
er
en
ce
will b
e
m
ea
s
u
r
ed
.
T
h
ese
d
if
f
er
e
n
ce
s
ca
n
b
e
e
m
p
lo
y
ed
in
t
h
e
r
ec
o
g
n
itio
n
d
ec
is
io
n
[
1
4
,
1
5
]
.
5.
P
RO
P
O
SE
D
T
E
CH
NIQU
E
AND
RE
SUL
T
S
B
y
m
ea
n
s
o
f
d
ata
m
in
in
g
an
d
J
av
a
Scr
ip
t,
f
o
r
v
o
ice
g
e
n
d
er
class
if
icatio
n
,
d
ata
s
et
is
co
llected
f
r
o
m
(
C
MU
_
AR
C
T
I
C
d
atab
ases
)
;
h
ttp
://fe
s
tv
o
x
.
o
r
g
/cm
u
_
a
r
ctic/cm
u
_
ar
ctic/cm
u
_
u
s
_
cl
b
_
ar
ctic/wa
v
/
.
Ma
k
e
p
r
e
-
em
p
h
asis
f
o
r
ea
ch
v
o
ice
s
am
p
le
in
th
e
d
atab
ase.
B
y
th
e
way
,
d
ata
s
et
is
2
2
7
0
(
1
1
3
8
m
ale,
1
1
3
2
f
em
ale
)
.
Data
m
in
in
g
s
tan
d
s
f
o
r
a
d
is
co
v
er
in
g
p
r
o
ce
s
s
f
o
r
p
atter
n
s
in
h
u
g
e
d
ata
s
ets
in
clu
d
in
g
m
et
h
o
d
s
at
th
e
in
ter
s
ec
tio
n
of
s
tatis
tics
,
m
ac
h
in
e
lear
n
in
g
an
d
d
atab
ase
s
y
s
tem
s
[
1
6
-
1
9
]
.
T
h
en
,
f
ea
tu
r
es
o
f
d
ata
a
r
e
s
p
lit
to
tr
ain
an
d
test
u
s
in
g
c
r
o
s
s
v
alid
atio
n
alg
o
r
ith
m
.
T
h
en
,
ex
tr
ac
t
MFC
C
f
ea
tu
r
es
f
o
r
ea
c
h
v
o
ice
a
n
d
ap
p
ly
v
ec
t
o
r
q
u
a
n
tizatio
n
o
n
ea
c
h
m
atr
ix
o
f
MFC
C
.
T
h
en
,
ex
t
r
ac
t
m
ea
n
(
s
u
m
o
f
f
ea
tu
r
es/t
h
eir
n
u
m
b
er
)
,
s
tan
d
ar
d
d
ev
iatio
n
(
STD)
,
ze
r
o
cr
o
s
s
in
g
(
Z
C
)
,
a
m
p
litu
d
e
(
AM
P
)
f
ea
t
u
r
es
f
o
r
ea
ch
v
o
ice.
Her
e
,
th
e
am
p
litu
d
e
s
tan
d
s
f
o
r
th
e
en
er
g
y
o
f
v
o
ice.
Feed
tr
ain
f
ea
tu
r
es
to
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
(
J
4
8
)
an
d
b
u
ild
class
if
ier
.
L
astl
y
,
f
ee
d
test
f
ea
tu
r
es
t
o
o
u
r
class
if
ier
to
class
if
y
v
o
ice
as
m
ale
o
r
f
e
m
ale.
T
h
e
p
r
o
p
o
s
ed
p
r
o
ce
d
u
r
e
is
d
e
p
icted
in
Fig
u
r
e
3.
J
4
8
alg
o
r
ith
m
s
tan
d
s
f
o
r
a
n
alg
o
r
ith
m
em
p
lo
y
e
d
f
o
r
p
r
o
d
u
cin
g
a
d
ec
is
io
n
tr
ee
estab
lis
h
ed
b
y
R
o
s
s
Qu
in
lan
.
I
t
h
as b
ee
n
a
n
e
x
ten
s
io
n
o
f
Q
u
in
lan
'
s
ea
r
lier
I
D3
alg
o
r
ith
m
.
T
h
e
d
ec
is
io
n
tr
e
es p
r
o
d
u
ce
d
b
y
J
4
8
alg
o
r
ith
m
ca
n
b
e
ad
o
p
ted
f
o
r
class
if
icatio
n
.
Acc
o
r
d
in
g
ly
,
J
4
8
alg
o
r
ith
m
is
f
r
eq
u
en
tly
r
e
f
er
r
ed
as
a
s
tatis
tical
class
if
ier
.
I
t
h
as
b
ee
n
p
o
wer
f
u
l
to
o
l
in
d
ata
m
in
i
n
g
[
2
0
]
.
T
h
e
d
etailed
ac
cu
r
ac
y
f
o
r
ea
ch
class
is
d
ep
icted
in
T
ab
le
1
.
T
ab
le
2
s
h
o
ws
th
e
co
m
p
ar
is
o
n
s
o
f
o
u
r
p
r
o
p
o
s
ed
v
o
ice
r
ec
o
g
n
itio
n
m
et
h
o
d
with
r
ep
o
r
ted
o
n
es
in
[
2
1
-
2
6
]
.
T
h
e
p
r
o
jecte
d
v
o
i
ce
r
ec
o
g
n
itio
n
m
eth
o
d
in
th
is
s
tu
d
y
h
as
b
etter
th
an
[
2
1
-
2
6
]
with
ac
c
u
r
ac
y
o
f
ab
o
u
t 1
0
0
%.
T
ab
le
1
.
R
esu
ltan
t a
cc
u
r
ac
y
f
o
r
ea
ch
class
W
e
i
g
h
t
e
d
A
v
g
TP R
a
t
e
F
P
R
a
t
e
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F
-
M
e
a
s
u
r
e
M
C
C
R
O
C
A
r
e
a
C
l
a
s
s
0
.
9
9
8
0
.
0
0
2
0
.
9
9
8
0
.
9
9
8
0
.
9
9
8
0
.
9
9
6
0
.
9
9
8
(
M
a
l
e
)
1
0
.
9
9
8
0
.
0
0
2
0
.
9
9
8
0
.
9
9
8
0
.
9
9
8
0
.
9
9
6
0
.
9
9
8
(
F
e
mal
e
)
2
0
.
9
9
8
0
.
0
0
2
0
.
9
9
8
0
.
9
9
8
0
.
9
9
8
0
.
9
9
6
0
.
9
9
8
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
,
Vo
l.
18
,
No
.
5
,
Octo
b
e
r
2
0
2
0
:
261
2
-
261
7
2616
T
ab
le
2
.
T
h
e
co
m
p
ar
is
o
n
s
o
f
p
r
o
p
o
s
ed
v
o
ice
r
ec
o
g
n
itio
n
m
et
h
o
d
with
r
e
p
o
r
te
d
ap
p
r
o
ac
h
es
in
[
2
1
-
26]
R
e
f
.
P
r
o
p
o
se
d
M
e
t
h
o
d
A
c
c
u
r
a
c
y
[
2
1
]
P
i
t
c
h
f
r
e
q
u
e
n
c
y
a
n
d
G
M
M
c
l
a
ss
i
f
i
e
r
9
8
.
6
5
%
[
2
2
]
4
c
l
a
ss
i
f
i
e
r
s i
n
v
o
l
v
i
n
g
G
M
M
,
v
e
c
t
o
r
q
u
a
n
t
i
z
a
t
i
o
n
(
V
Q
)
,
mu
l
t
i
l
a
y
e
r
p
e
r
c
e
p
t
r
o
n
(
M
LP)
,
a
n
d
l
e
a
r
n
i
n
g
v
e
c
t
o
r
q
u
a
n
t
i
z
a
t
i
o
n
(
LV
Q
)
9
6
.
4
%
[
2
3
]
j
o
i
n
t
e
st
i
ma
t
e
d
v
o
i
c
e
a
c
o
u
s
t
i
c
l
e
v
e
l
o
f
5
d
i
v
e
r
se
a
p
p
r
o
a
c
h
e
s
i
n
t
o
s
i
n
g
l
e
sco
r
e
l
e
v
e
l
8
1
.
7
%
[
2
4
]
F
u
si
o
n
sc
o
r
e
o
f
7
s
u
b
sy
s
t
e
ms
.
T
h
e
f
e
a
t
u
r
e
v
e
c
t
o
r
s
i
n
c
l
u
d
e
t
h
e
M
F
C
C
,
P
LP,
a
n
d
p
r
o
s
o
d
i
c
o
n
3
c
l
a
ssi
f
i
e
r
s
o
f
G
M
M
,
S
V
M
,
a
n
d
G
M
M
-
SV
-
b
a
se
d
S
V
M
c
o
m
b
i
n
e
d
a
t
t
h
e
s
c
o
r
e
l
e
v
e
l
9
0
.
4
[
2
5
]
D
e
c
i
s
i
o
n
t
r
e
e
(
D
T)
a
n
d
S
V
M
w
i
t
h
t
h
e
M
F
C
C
f
e
a
t
u
r
e
,
o
n
a
n
i
so
l
a
t
e
d
C
o
r
p
u
s
f
o
r
c
l
a
ssi
f
y
i
n
g
g
e
n
d
e
r
v
o
i
c
e
9
3
.
1
6
%
a
n
d
9
1
.
4
5
%
f
o
r
M
F
C
C
-
S
V
M
a
n
d
M
F
C
C
-
DT
[
2
6
]
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
a
n
d
D
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
s (D
N
N
s)
f
o
r
M
F
C
C
e
n
h
a
n
c
e
me
n
t
5
8
.
9
8
%
a
n
d
5
6
.
1
3
%
,
e
v
a
l
u
a
t
e
d
o
n
D
N
N
a
n
d
I
-
V
e
c
t
o
r
c
l
a
s
si
f
i
e
r
s
Th
i
s
st
u
d
y
M
F
C
C
,
V
Q
,
a
n
d
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
m (J
4
8
)
9
9
.
8
%
Fig
u
r
e
3
.
Pro
p
o
s
ed
p
r
o
ce
d
u
r
e
f
o
r
v
o
ice
g
en
d
er
class
if
icatio
n
in
th
is
s
tu
d
y
6.
CO
NCLU
SI
O
N
I
n
th
is
p
ap
e
r
,
v
o
ice
g
en
d
er
cl
ass
if
icatio
n
h
as
b
ee
n
im
p
lem
e
n
ted
b
ased
o
n
MFC
C
,
VQ,
an
d
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
(
J
4
8
)
.
T
h
is
class
if
icatio
n
s
y
s
tem
is
te
s
ted
o
v
er
v
o
ice
s
am
p
le
s
o
f
2
2
7
0
c
eleb
r
ities
in
clu
d
in
g
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
Gen
d
er v
o
ice
cla
s
s
ifica
tio
n
w
ith
h
u
g
e
a
cc
u
r
a
cy
r
a
te
(
Mu
s
ta
f
a
S
a
h
i
b
S
h
a
r
ee
f
)
2617
1
1
3
8
m
ales
an
d
1
1
3
2
f
em
al
es.
T
h
e
class
if
icatio
n
ac
cu
r
a
cy
o
f
ab
o
u
t
1
0
0
%
is
ac
h
iev
ed
b
y
th
e
p
r
o
p
o
s
ed
class
if
icatio
n
tech
n
iq
u
e
th
at
is
h
ig
h
er
t
h
an
m
a
n
y
r
e
p
o
r
ted
tecn
iq
u
es in
th
e
liter
atu
r
e.
RE
F
E
R
E
NC
E
S
[1
]
S
u
k
h
o
sta
t
L.
,
a
n
d
Im
a
m
v
e
rd
iy
e
v
Y.
,
“
A
c
o
m
p
a
ra
ti
v
e
a
n
a
ly
sis
o
f
p
it
c
h
d
e
tec
ti
o
n
m
e
th
o
d
s
u
n
d
e
r
th
e
in
flu
e
n
c
e
o
f
d
iffere
n
t
n
o
ise
c
o
n
d
it
io
n
s
,”
J
o
u
r
n
a
l
o
f
v
o
ice
,
v
o
l.
29
,
n
o
.
4
,
p
p
.
4
1
0
-
4
1
7
,
2
0
1
5
.
[2
]
Ha
u
tam
ä
k
i
R.
G
.
,
S
a
h
id
u
l
lah
M
.
,
Ha
u
tam
ä
k
i
V.,
Kin
n
u
n
e
n
T.
,
“
Ac
o
u
stica
l
a
n
d
p
e
rc
e
p
tu
a
l
st
u
d
y
o
f
v
o
ice
d
isg
u
ise
b
y
a
g
e
m
o
d
ifi
c
a
ti
o
n
i
n
sp
e
a
k
e
r
v
e
rifi
c
a
ti
o
n
,”
S
p
e
e
c
h
Co
mm
u
n
ica
t
io
n
,
v
o
l.
95
,
p
p
.
1
-
15
,
2
0
1
7
.
[3
]
Ei
c
h
h
o
rn
,
Ju
li
e
Trau
b
,
e
t
a
l.
"
Ef
fe
c
ts
o
f
a
g
i
n
g
o
n
v
o
c
a
l
fu
n
d
a
m
e
n
tal
fre
q
u
e
n
c
y
a
n
d
v
o
we
l
fo
rm
a
n
ts
i
n
m
e
n
a
n
d
wo
m
e
n
,
"
J
o
u
rn
a
l
o
f
V
o
ice
,
v
o
l.
3
2
,
n
o
.
5
,
2
0
1
8
.
[4
]
P
h
o
o
p
h
u
a
n
g
p
a
ir
o
j,
Ro
n
g
,
a
n
d
S
u
k
a
n
y
a
P
h
o
n
g
su
p
h
a
p
,
"
Tw
o
-
S
tag
e
G
e
n
d
e
r
Id
e
n
ti
fica
ti
o
n
Us
i
n
g
P
it
c
h
F
re
q
u
e
n
c
ies
,
M
F
CCs
a
n
d
HMM
s
,
"
2
0
1
5
IEE
E
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
S
y
ste
ms
,
M
a
n
,
a
n
d
Cy
b
e
rn
e
ti
c
s
,
IEE
E
,
2
0
1
5
.
[5
]
Na
sr
M
.
A.
,
A
b
d
-
E
ln
a
b
y
M
.
,
El
-
F
ish
a
wy
A
.
S
.
,
El
-
Ra
b
a
ie
S
.
,
Ab
d
E
l
-
S
a
m
ie
F
.
E.
,
“
Eff
icie
n
t
imp
lem
e
n
tatio
n
o
f
a
d
a
p
ti
v
e
wie
n
e
r
fil
ter
fo
r
p
it
c
h
d
e
tec
ti
o
n
fro
m
n
o
isy
sp
e
e
c
h
sig
n
a
ls
,”
M
e
n
o
u
fi
a
J
o
u
r
n
a
l
o
f
El
e
c
tro
n
ic
En
g
in
e
e
rin
g
Res
e
a
rc
h
,
v
o
l.
27
,
n
o
.
1
,
p
p
.
1
0
9
-
126
,
2
0
1
8
.
[6
]
Eri
c
sd
o
tt
e
r
C
.
,
Eri
c
ss
o
n
A
.
M
.,
“
G
e
n
d
e
r
d
iffere
n
c
e
s in
v
o
we
l
d
u
ra
t
io
n
i
n
re
a
d
S
we
d
is
h
:
P
re
li
m
in
a
ry
re
su
lt
s W
o
rk
i
n
g
P
a
p
e
rs
,”
L
u
n
d
U
n
ive
rs
it
y
De
p
a
rt
me
n
t
o
f
L
in
g
u
ist
ics
,
p
p
.
34
-
37
,
2
0
0
1
.
[7
]
Wh
it
e
sid
e
S
.
P
.,
“
Tem
p
o
ra
l
-
Ba
se
d
Ac
o
u
stic
-
P
h
o
n
e
ti
c
P
a
tt
e
r
n
s
i
n
Re
a
d
S
p
e
e
c
h
:
S
o
m
e
Ev
i
d
e
n
c
e
f
o
r
S
p
e
a
k
e
r
S
e
x
Diffe
re
n
c
e
s,
”
J
In
ter
n
a
ti
o
n
a
l
Ph
o
n
e
ti
c
Asso
c
ia
ti
o
n
,
v
o
l.
26
,
p
p
.
23
-
40
,
1
9
9
6
.
[8
]
He
n
to
n
C
.
G
.,
“
F
a
c
t
a
n
d
ficti
o
n
in
th
e
d
e
sc
rip
ti
o
n
o
f
m
a
le
a
n
d
fe
m
a
le
p
it
c
h
,”
L
a
n
g
u
a
g
e
a
n
d
Co
mm
u
n
ica
ti
o
n
,
v
o
l.
9
,
p
p
.
2
9
9
-
311
,
1
9
8
9
.
[9
]
Bish
o
p
J
.,
a
n
d
Ke
a
ti
n
g
P
.,
“
P
e
rc
e
p
ti
o
n
o
f
p
it
c
h
lo
c
a
ti
o
n
wit
h
i
n
a
sp
e
a
k
e
r’s
ra
n
g
e
:
F
u
n
d
a
m
e
n
tal
F
r
e
q
u
e
n
c
y
,
v
o
ice
q
u
a
li
t
y
a
n
d
sp
e
a
k
e
r
se
x
,”
T
h
e
J
o
u
rn
a
l
o
f
t
h
e
Aco
u
stica
l
S
o
c
iety
o
f
Ame
ric
a
,
v
o
l.
32
,
n
o
.
2
,
p
p
.
1
1
0
0
-
1
1
1
2
,
2
0
1
2
.
[1
0
]
S
m
i
th
D
.
R
.,
a
n
d
P
a
tt
e
rso
n
R
.
D
.,
“
Th
e
in
tera
c
ti
o
n
o
f
g
lo
tt
a
l
-
p
u
lse
r
a
te
a
n
d
v
o
c
a
l
-
trac
t
len
g
t
h
i
n
j
u
d
g
m
e
n
ts
o
f
s
p
e
a
k
e
r
siz
e
,
se
x
,
a
n
d
a
g
e
,”
T
h
e
J
o
u
rn
a
l
o
f
th
e
Aco
u
stica
l
S
o
c
iety
o
f
Ame
ric
a
,
v
o
l.
1
1
8
,
n
o
.
5
,
p
p
.
3
1
7
7
-
3
1
8
6
,
2
0
0
5
.
[
1
1
]
M
u
h
a
m
m
a
d
G
.
,
A
l
S
u
l
a
i
m
a
n
M
.
,
M
a
h
m
o
o
d
A
.,
a
n
d
A
l
i
Z
.,
“
A
u
t
o
m
a
t
i
c
v
o
i
c
e
d
i
s
o
r
d
e
r
c
l
a
s
s
i
f
i
c
a
t
i
o
n
u
s
i
n
g
v
o
w
e
l
f
o
r
m
a
n
t
s
,
”
2
0
1
1
P
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
E
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
M
u
l
t
i
m
e
d
i
a
a
n
d
E
x
p
o
(
I
C
M
E
'
1
1
)
,
p
p
.
1
-
6
,
2
0
1
1
[1
2
]
G
a
ik
wa
d
S
.
,
G
a
wa
li
B
.
,
a
n
d
M
e
h
ro
tra
S
.
C
.,
“
G
e
n
d
e
r
id
e
n
t
ifi
c
a
ti
o
n
u
si
n
g
S
VM
wit
h
c
o
m
b
in
a
t
io
n
o
f
M
F
CC,
”
Ad
v
a
n
c
e
s i
n
Co
m
p
u
t
a
ti
o
n
a
l
Res
e
a
rc
h
,
v
o
l.
4
,
p
p
.
69
-
73
,
2
0
1
2
.
[1
3
]
Zen
g
Y
.
M
.
,
W
u
Z
.
Y
.
,
F
a
l
k
T
.,
a
n
d
Ch
a
n
W
.
Y
.,
“
Ro
b
u
st
G
M
M
b
a
se
d
g
e
n
d
e
r
c
las
sifica
ti
o
n
u
sin
g
p
it
c
h
a
n
d
RAST
A
-
P
LP
p
a
ra
m
e
ters
o
f
sp
e
e
c
h
,”
2
0
0
6
Pro
c
e
e
d
in
g
s
o
f
t
h
e
In
t
e
rn
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
a
n
d
Cy
b
e
rn
e
ti
c
s
,
p
p
.
3
3
7
6
-
3
3
7
9
,
2
0
0
6
.
[1
4
]
P
a
tel
Ka
sh
y
a
p
,
R.
K.
P
ra
sa
d
,
"
S
p
e
e
c
h
re
c
o
g
n
it
i
o
n
a
n
d
v
e
rifi
c
a
ti
o
n
u
sin
g
M
F
CC
&
VQ
,
"
I
n
t.
J
.
E
me
rg
.
S
c
i.
En
g
.
,
v
o
l.
1
,
n
o
.
7
,
p
p
.
3
3
3
-
37,
2
0
1
3
.
[1
5
]
Alk
h
a
wa
ld
e
h
Ra
m
i
S
.
,
"
DG
R:
G
e
n
d
e
r
Re
c
o
g
n
it
i
o
n
o
f
H
u
m
a
n
S
p
e
e
c
h
Us
in
g
O
n
e
-
Dim
e
n
sio
n
a
l
C
o
n
v
e
n
ti
o
n
a
l
Ne
u
ra
l
Ne
two
rk
,
"
S
c
ien
ti
f
ic P
ro
g
ra
mm
in
g
,
v
o
l.
2
0
1
9
,
p
p
.
1
1
2
,
2
0
1
9
.
[1
6
]
Ro
ig
e
r
R
.
J.
,
“
Da
ta m
in
in
g
:
a
t
u
t
o
rial
-
b
a
se
d
p
rime
r
,”
CRC
p
re
ss
;
2
0
1
7
.
[1
7
]
Du
tt
A
.
,
Ism
a
il
M
.
A
.
,
He
ra
wa
n
T.
,
“
A
sy
ste
m
a
ti
c
re
v
iew
o
n
e
d
u
c
a
ti
o
n
a
l
d
a
ta
m
in
in
g
,”
IEE
E
Acc
e
ss
,
v
o
l.
17
,
n
o
.
5
,
p
p
.
1
5
9
9
1
-
6
0
0
5
,
2
0
1
7
.
[1
8
]
S
a
m
m
u
t,
Clau
d
e
,
a
n
d
G
e
o
ffre
y
I.
Web
b
,
“
E
n
c
y
c
lo
p
e
d
ia
o
f
m
a
c
h
in
e
lea
rn
in
g
a
n
d
d
a
ta
m
i
n
in
g
,”
S
p
rin
g
e
r
Pu
b
li
sh
i
n
g
Co
mp
a
n
y
,
I
n
c
o
rp
o
ra
ted
,
2
0
1
7
.
[1
9
]
G
e
Zh
iq
ian
g
,
Zh
i
h
u
a
n
S
o
n
g
,
S
tev
e
n
X.
Din
g
,
a
n
d
B
iao
Hu
a
n
g
,
"
D
a
ta
m
in
in
g
a
n
d
a
n
a
ly
t
ics
in
th
e
p
r
o
c
e
ss
in
d
u
str
y
:
Th
e
ro
le
o
f
m
a
c
h
in
e
lea
rn
i
n
g
,
"
I
E
EE
Acc
e
ss
,
v
o
l.
5
,
p
p
.
2
0
5
9
0
-
2
0
6
1
6
,
2
0
1
7
.
[2
0
]
Bh
a
rg
a
v
a
N.
,
S
h
a
rm
a
G
.
,
Bh
a
r
g
a
v
a
R.
,
M
a
th
u
ria
M
.
,
“
De
c
isio
n
tree
a
n
a
ly
sis
o
n
j
4
8
a
lg
o
rit
h
m
fo
r
d
a
ta
m
i
n
in
g
,
”
Pro
c
e
e
d
in
g
s
o
f
I
n
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
A
d
v
a
n
c
e
d
Res
e
a
rc
h
i
n
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
S
o
ft
w
a
re
En
g
in
e
e
rin
g
,
2
0
1
3
.
[2
1
]
Y.
Hu
,
D.
Wu
,
a
n
d
A
.
Nu
c
c
i,
“
P
it
c
h
-
b
a
se
d
g
e
n
d
e
r
id
e
n
ti
fica
ti
o
n
with
tw
o
-
sta
g
e
c
las
sifica
ti
o
n
,
”
S
e
c
u
rit
y
a
n
d
Co
m
m
u
n
ica
ti
o
n
Ne
two
rk
s
,
v
o
l.
5
,
n
o
.
2
,
p
p
.
2
1
1
-
2
2
5
,
2
0
1
2
.
[
2
2
]
R
.
D
j
e
m
i
l
i
,
H
.
B
o
u
r
o
u
b
a
,
a
n
d
M
.
C
.
A
.
K
o
r
b
a
,
“
A
s
p
e
e
c
h
s
i
g
n
a
l
-
b
a
s
e
d
g
e
n
d
e
r
i
d
e
n
t
i
f
i
c
a
t
i
o
n
s
y
s
t
e
m
u
s
i
n
g
f
o
u
r
c
l
a
s
s
i
f
i
e
r
s
,
”
P
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
2
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
M
u
l
t
i
m
e
d
i
a
C
o
m
p
u
t
i
n
g
a
n
d
S
y
s
t
e
m
s
,
p
p
.
1
8
4
-
1
8
7
,
2
0
1
2
.
[2
3
]
M
.
Li
,
K.
J.
Ha
n
,
a
n
d
S
.
Na
ra
y
a
n
a
n
,
“
Au
t
o
m
a
ti
c
sp
e
a
k
e
r
a
g
e
a
n
d
g
e
n
d
e
r
re
c
o
g
n
it
i
o
n
u
sin
g
a
c
o
u
sti
c
a
n
d
p
ro
s
o
d
i
c
lev
e
l
in
f
o
rm
a
ti
o
n
fu
si
o
n
,
”
C
o
mp
u
ter
S
p
e
e
c
h
&
L
a
n
g
u
a
g
e
,
v
o
l
.
2
7
,
n
o
.
1
,
p
p
.
1
5
1
-
1
6
7
,
2
0
1
3
.
[2
4
]
E.
Yu
c
e
so
y
a
n
d
V.
V.
Na
b
i
y
e
v
,
“
A
n
e
w
a
p
p
ro
a
c
h
wit
h
sc
o
re
lev
e
l
f
u
sio
n
f
o
r
t
h
e
c
las
sifica
ti
o
n
o
f
a
sp
e
a
k
e
r
a
g
e
a
n
d
g
e
n
d
e
r,
”
C
o
mp
u
ter
s &
El
e
c
trica
l
En
g
i
n
e
e
rin
g
,
v
o
l.
5
3
,
p
p
.
2
9
-
3
9
,
2
0
1
6
.
[
2
5
]
M.
W
.
L
e
e
a
n
d
K
.
C
.
K
w
a
k
,
“
P
e
r
f
o
r
m
a
n
c
e
c
o
m
p
a
r
i
s
o
n
o
f
g
e
n
d
e
r
a
n
d
a
g
e
g
r
o
u
p
r
e
c
o
g
n
i
t
i
o
n
f
o
r
h
u
m
a
n
-
r
o
b
o
t
i
n
t
e
r
a
c
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
A
d
v
a
n
c
e
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
3
,
n
o
.
1
2
,
p
p
.
2
0
7
-
2
1
1
,
2012.
[2
6
]
Z.
Qa
wa
q
n
e
h
,
A.
A.
M
a
ll
o
u
h
,
a
n
d
B.
D.
Ba
rk
a
n
a
,
“
De
e
p
n
e
u
ra
l
n
e
two
rk
fra
m
e
wo
rk
a
n
d
tran
sfo
rm
e
d
M
F
CCs
f
o
r
sp
e
a
k
e
r’s ag
e
a
n
d
g
e
n
d
e
r
c
las
sific
a
ti
o
n
,
”
K
n
o
wle
d
g
e
-
Ba
se
d
S
y
ste
ms
,
v
o
l.
1
1
5
,
p
p
.
5
-
1
4
,
2
0
1
7
.
Evaluation Warning : The document was created with Spire.PDF for Python.