I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
6
,
No
.
6
,
Decem
b
er
201
6
,
p
p
.
276
6
~
2
7
7
2
I
SS
N:
2088
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
6i
6
.
9
8
4
4
2766
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
jo
u
r
n
a
l.c
o
m/o
n
lin
e/in
d
ex
.
p
h
p
/I
JE
C
E
O
n
Usa
ble Spee
c
h Det
ection
by
Li
nea
r Multi
-
Sca
le
Deco
m
po
sitio
n f
o
r Spea
k
er
Identi
fi
ca
tion
Wa
j
di G
heza
iel
1
,
A
m
el
B
en
Sli
m
a
ne
2
,
E
zz
e
din
e
B
en
B
ra
iek
3
1
CERE
P
-
ENS
I
T
,
Un
iv
e
rsity
of
Tu
n
is,
T
u
n
is,
T
u
n
isia
2
,3
ENS
I,
U
n
iv
e
rsity
of
M
a
n
o
u
b
a
,
M
a
n
o
u
b
a
,
T
u
n
isia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Dec
3
1
,
2
0
1
5
R
ev
i
s
ed
Ma
r
1
5
,
2
0
1
6
A
cc
ep
ted
Mar
29
,
2
0
1
6
Us
a
b
le
sp
e
e
c
h
is
a
n
o
v
e
l
c
o
n
c
e
p
t
o
f
p
ro
c
e
ss
in
g
c
o
-
c
h
a
n
n
e
l
sp
e
e
c
h
d
a
ta.
It
is
p
ro
p
o
se
d
to
e
x
trac
t
m
in
im
a
ll
y
c
o
rru
p
ted
sp
e
e
c
h
t
h
a
t
is
c
o
n
sid
e
re
d
u
se
f
u
l
f
o
r
v
a
rio
u
s
sp
e
e
c
h
p
ro
c
e
ss
in
g
s
y
ste
m
s.
In
th
is
p
a
p
e
r,
w
e
a
re
in
tere
s
ted
f
o
r
c
o
-
c
h
a
n
n
e
l
sp
e
a
k
e
r
id
e
n
ti
f
ica
ti
o
n
(S
ID).
W
e
e
m
p
lo
y
a
n
e
w
p
ro
p
o
se
d
u
sa
b
le
sp
e
e
c
h
e
x
trac
ti
o
n
m
e
th
o
d
b
a
se
d
o
n
th
e
p
it
c
h
in
f
o
rm
a
ti
o
n
o
b
tai
n
e
d
f
ro
m
li
n
e
a
r
m
u
lt
i
-
sc
a
le
d
e
c
o
m
p
o
siti
o
n
b
y
d
isc
re
te
wa
v
e
l
e
t
tran
sf
o
r
m
.
T
h
e
id
e
a
is
to
re
tain
t
h
e
sp
e
e
c
h
se
g
m
e
n
ts
th
a
t
h
a
v
e
o
n
ly
o
n
e
p
it
c
h
d
e
tec
ted
a
n
d
re
m
o
v
e
th
e
o
th
e
rs.
De
tec
ted
Us
a
b
le
sp
e
e
c
h
wa
s
u
se
d
a
s
in
p
u
t
f
o
r
sp
e
a
k
e
r
id
e
n
ti
f
ica
ti
o
n
sy
ste
m
.
T
h
e
s
y
ste
m
is
e
v
a
lu
a
ted
o
n
c
o
-
c
h
a
n
n
e
l
s
p
e
e
c
h
a
n
d
re
su
lt
s
sh
o
w
a
sig
n
i
f
ic
a
n
t
i
m
p
ro
v
e
m
e
n
t
a
c
ro
ss
v
a
rio
u
s
t
a
rg
e
t
to
In
terf
e
re
r
Ra
ti
o
(T
IR)
f
o
r
sp
e
a
k
e
r
id
e
n
ti
f
ica
ti
o
n
sy
ste
m
.
K
ey
w
o
r
d
:
C
o
-
c
h
an
n
el
s
p
ee
c
h
Dis
cr
ete
w
av
ele
t tr
an
s
f
o
r
m
Mu
lti
-
s
ca
le
d
ec
o
m
p
o
s
itio
n
Sp
ea
k
er
id
en
ti
f
icatio
n
Usab
le
s
p
ee
ch
Co
p
y
rig
h
t
©
2
0
1
6
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts
re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
W
ajd
i G
h
ez
aiel
,
C
E
R
E
P
-
E
NSI
T
,
Un
i
v
er
s
it
y
o
f
T
u
n
i
s
,
T
u
n
is
,
T
u
n
is
ia.
E
m
ail:
w
aj
d
i.g
h
ez
aie
l@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Sp
ee
ch
ca
n
b
e
d
is
to
r
t
ed
b
y
m
an
y
k
i
n
d
s
o
f
i
n
ter
f
er
en
ce
s
.
I
n
ter
f
er
in
g
s
i
g
n
al
ca
n
b
e
s
tatio
n
ar
y
o
r
n
o
s
tatio
n
ar
y
s
i
g
n
al.
Statio
n
ar
y
n
o
is
e
ca
n
b
e
d
ea
lt
w
it
h
d
en
o
is
i
n
g
a
n
d
n
o
i
s
e
r
ed
u
ctio
n
tec
h
n
i
q
u
es;
w
h
er
ea
s
n
o
n
s
tatio
n
ar
y
n
o
i
s
e
is
ca
u
s
ed
b
y
an
o
th
er
s
p
e
ec
h
f
r
o
m
a
d
i
f
f
er
e
n
t
s
p
ea
k
er
.
Su
c
h
i
n
ter
f
er
en
ce
is
f
r
eq
u
en
t
a
n
d
t
h
e
co
r
r
u
p
ted
s
p
ee
ch
is
k
n
o
w
n
a
s
co
-
ch
a
n
n
e
l
s
p
ee
ch
[
1
]
.
M
a
n
y
s
p
ee
c
h
p
r
o
ce
s
s
in
g
tech
n
iq
u
es
ar
e
p
lag
u
ed
f
o
r
s
u
c
h
in
ter
f
er
en
ce
s
.
T
r
a
d
itio
n
al
ap
p
r
o
ac
h
to
co
-
ch
an
n
e
l
s
p
ee
ch
is
to
atte
m
p
t
to
ex
tr
ac
t
th
e
s
p
ee
ch
o
f
th
e
s
p
ea
k
er
o
f
in
ter
est
(
tar
g
et
s
p
ee
ch
)
f
r
o
m
o
th
er
(
in
ter
f
er
in
g
)
s
p
ee
ch
.
Usab
le
s
p
ee
ch
ex
tr
ac
tio
n
is
a
n
o
v
el
co
n
ce
p
t
o
f
p
r
o
ce
s
s
in
g
d
e
g
r
ad
ed
s
p
ee
ch
d
ata.
T
h
e
id
ea
o
f
u
s
ab
le
s
p
ee
ch
i
s
to
id
en
ti
f
y
a
n
d
to
ex
tr
ac
t
p
o
r
tio
n
s
o
f
d
eg
r
ad
ed
s
p
ee
ch
th
at
ar
e
co
n
s
id
er
ed
u
s
e
f
u
l f
o
r
v
ar
io
u
s
s
p
e
ec
h
p
r
o
ce
s
s
in
g
s
y
s
te
m
s
.
Sp
ea
k
er
id
en
ti
f
icatio
n
s
y
s
te
m
[2
-
4
]
n
ee
d
s
p
o
r
tio
n
s
o
f
s
p
ee
ch
th
at
co
n
tai
n
s
p
ea
k
er
ch
ar
ac
ter
is
tics
,
w
h
ic
h
a
r
e
u
n
iq
u
e
to
th
e
in
d
i
v
id
u
al
s
p
ea
k
er
s
,
class
if
iab
le
an
d
lo
n
g
en
o
u
g
h
f
o
r
th
e
s
y
s
te
m
s
to
m
a
k
e
t
h
e
d
ec
is
io
n
[
1
]
.
T
h
ese
p
o
r
tio
n
s
o
f
s
p
ee
ch
ar
e
ter
m
ed
as
u
s
ab
le
s
p
ee
ch
an
d
d
ef
in
ed
as
co
n
s
ec
u
ti
v
e
f
r
a
m
e
s
o
f
s
p
ee
ch
th
a
t
ar
e
m
i
n
i
m
a
ll
y
co
r
r
u
p
ted
b
y
in
ter
f
er
i
n
g
s
p
ee
ch
.
Du
e
to
th
e
n
at
u
r
e
o
f
h
u
m
a
n
v
o
ice,
a
s
p
ee
ch
u
tter
an
ce
co
n
tai
n
s
v
o
iced
p
ar
ts
,
u
n
v
o
iced
p
ar
ts
an
d
s
ile
n
ce
;
af
ter
m
i
x
i
n
g
th
e
t
w
o
s
p
ee
c
h
s
ig
n
al
s
,
t
h
er
e
ar
e
s
eg
m
e
n
ts
o
f
t
h
e
co
-
c
h
an
n
el
s
p
ee
ch
th
at
co
n
tai
n
o
n
l
y
o
n
e
s
p
ea
k
er
‟
s
v
o
iced
p
ar
t
o
r
o
n
e
s
p
ea
k
er
‟
s
v
o
iced
p
ar
t
p
lu
s
an
o
t
h
er
s
p
ea
k
er
‟
s
u
n
v
o
ic
ed
p
ar
t,
th
e
latter
u
s
u
a
ll
y
h
a
v
i
n
g
m
u
ch
lo
w
er
e
n
er
g
y
.
Yan
to
r
n
o
[
5
]
p
e
r
f
o
r
m
ed
a
s
tu
d
y
o
n
co
-
c
h
an
n
el
s
p
ee
c
h
a
n
d
co
n
clu
d
ed
th
at
t
h
e
T
ar
g
et
-
to
-
I
n
ter
f
er
er
R
atio
(
T
I
R
)
w
a
s
a
g
o
o
d
m
ea
s
u
r
e
to
q
u
an
ti
f
y
u
s
ab
ilit
y
f
o
r
s
p
ea
k
er
id
en
ti
f
i
ca
tio
n
.
U
s
ab
le
s
eg
m
e
n
t
ex
tr
ac
tio
n
i
s
b
ased
o
n
a
p
o
w
e
r
r
atio
o
f
th
e
tar
g
et
s
p
ee
ch
to
th
e
in
ter
f
er
in
g
s
p
ee
ch
.
T
h
is
r
atio
is
ex
p
r
ess
ed
as
T
I
R
(
T
a
r
g
et
to
I
n
ter
f
er
er
R
atio
,
in
d
B
)
.
T
h
e
r
atio
ca
n
b
e
ex
p
r
es
s
ed
f
o
r
e
n
tire
u
tt
er
an
ce
s
o
r
i
n
d
iv
id
u
a
l
f
r
a
m
e
s
o
f
s
p
ee
c
h
.
Fo
r
u
s
ab
ilit
y
,
p
r
ev
io
u
s
e
x
p
er
i
m
e
n
tatio
n
h
as
s
h
o
w
n
t
h
at
f
o
r
f
r
a
m
es
ab
o
v
e
2
0
d
B
T
I
R
is
co
n
s
id
er
ed
u
s
ab
le,
a
n
d
th
at
lo
w
er
2
0
d
B
T
I
R
is
co
n
s
id
er
ed
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
On
Usa
b
le
S
p
ee
ch
Dete
ctio
n
b
y
Lin
ea
r
Mu
lti
-
S
ca
le
Dec
o
mp
o
s
itio
n
fo
r
S
I
D
(
W
a
jd
i G
h
eza
i
el
)
2767
u
n
u
s
ab
le
s
e
g
m
e
n
t
s
.
T
h
e
co
n
c
ep
t
o
f
“
u
s
ab
le”
s
e
g
m
en
t
s
r
eli
es
o
n
t
h
e
f
ac
t
t
h
at
f
o
r
an
y
g
i
v
en
ti
m
e
f
r
a
m
e,
t
h
e
en
er
g
y
o
f
ea
ch
s
p
ea
k
er
m
a
y
b
e
d
if
f
er
e
n
t.
T
h
e
u
s
ab
le
s
p
ee
ch
co
n
ce
p
t
tak
e
s
ad
v
a
n
ta
g
e
o
f
t
h
e
s
it
u
atio
n
w
h
e
n
th
e
en
er
g
y
o
f
th
e
p
r
i
m
ar
y
s
p
ea
k
er
is
m
u
c
h
g
r
ea
ter
th
a
n
t
h
e
e
n
er
g
y
o
f
t
h
e
in
ter
f
er
i
n
g
s
p
ea
k
er
f
o
r
a
g
i
v
en
f
r
a
m
e.
D
if
f
er
e
n
t
cr
iter
ia
ar
e
d
ev
elo
p
ed
to
ex
tr
ac
t
u
s
ab
le
s
p
ee
ch
in
co
-
c
h
an
n
el
s
p
ee
ch
[
6
-
7
]
.
C
r
iter
ia
s
u
ch
f
r
a
m
e
-
le
v
el
T
I
R
o
r
s
p
ec
tr
al
au
to
co
r
r
elatio
n
r
atio
.
T
h
ese
s
t
u
d
ies
f
i
n
d
t
h
at
v
o
iced
s
e
g
m
en
t
s
co
n
tain
m
o
s
t
o
f
t
h
e
in
f
o
r
m
atio
n
f
o
r
SID
,
an
d
ac
co
r
d
in
g
to
th
ese
cr
iter
ia,
a
s
ig
n
if
ican
t
a
m
o
u
n
t
o
f
co
-
c
h
an
n
el
s
p
ee
ch
ca
n
b
e
co
n
s
id
er
ed
u
s
ab
le
f
o
r
SID
.
Fra
m
e
T
I
R
s
ar
e
ea
s
il
y
ca
lcu
late
d
w
ith
p
r
e
m
i
x
i
n
g
s
p
ee
c
h
u
tte
r
an
ce
s
,
a
n
d
u
s
ab
le
s
p
ee
ch
e
x
tr
ac
ted
b
ased
o
n
a
T
I
R
th
r
esh
o
ld
r
etai
n
s
f
r
a
m
e
s
w
h
er
e
tar
g
e
t
s
p
ea
k
er
is
m
u
ch
s
tr
o
n
g
er
i
n
ter
m
s
o
f
o
v
er
all
en
er
g
y
th
a
n
th
e
o
t
h
er
.
Sp
ec
tr
al
au
to
co
r
r
elatio
n
r
atio
esti
m
ate
s
th
e
r
atio
b
et
w
ee
n
d
o
m
i
n
a
n
t
p
ea
k
an
d
v
alle
y
in
a
u
to
co
r
r
elatio
n
o
f
a
s
p
ec
tr
al
f
r
a
m
e.
T
h
is
r
at
io
is
u
s
ed
to
d
eter
m
i
n
e
w
h
eth
er
a
f
r
a
m
e
i
s
u
s
ab
le,
m
ea
n
in
g
t
h
e
s
p
ec
tr
u
m
is
w
e
ll
s
tr
u
c
tu
r
ed
(
s
i
n
g
le
-
s
p
ea
k
er
s
p
ee
ch
)
,
o
r
n
o
t.
T
h
is
ap
p
r
o
a
ch
is
s
i
m
p
le
an
d
ef
f
ec
tiv
e
an
d
s
h
o
w
s
a
s
u
b
s
tan
tial
i
m
p
r
o
v
e
m
e
n
t
i
n
SID
p
er
f
o
r
m
a
n
ce
.
He
n
ce
,
a
n
u
m
b
er
o
f
m
et
h
o
d
s
f
o
r
u
s
ab
le
s
p
ee
ch
d
etec
tio
n
w
h
ich
r
ef
er
t
o
th
e
T
I
R
h
a
v
e
b
ee
n
d
ev
elo
p
e
d
an
d
s
t
u
d
ied
u
n
d
er
co
-
c
h
an
n
el
co
n
d
itio
n
[
8
]
[
9
]
.
I
n
th
ese
m
eth
o
d
s
,
u
s
ab
le
s
p
ee
ch
f
r
a
m
e
s
ar
e
co
m
p
o
s
ed
o
f
v
o
iced
s
p
ee
ch
.
I
n
[
8
]
,
th
e
Sp
ec
tr
al
A
u
to
co
r
r
elatio
n
R
atio
m
et
h
o
d
w
a
s
d
ev
elo
p
ed
to
d
etec
t
u
s
ab
le
s
p
ee
ch
s
eg
m
en
ts
.
T
h
is
tak
e
s
ad
v
an
tag
e
o
f
th
e
s
tr
u
ctu
r
e
o
f
v
o
iced
s
p
ee
c
h
i
n
t
h
e
f
r
eq
u
e
n
c
y
d
o
m
ai
n
.
I
n
[
9
]
,
t
h
e
P
ea
k
d
if
f
er
e
n
ce
a
u
to
co
r
r
elatio
n
o
f
w
a
v
elet
tr
an
s
f
o
r
m
(
P
DA
W
T
)
m
et
h
o
d
is
ap
p
lied
in
o
r
d
er
to
d
etec
t
p
itch
in
f
o
r
m
atio
n
i
n
u
s
ab
le
s
p
ee
ch
.
T
h
is
m
et
h
o
d
ap
p
lies
au
to
co
r
r
elatio
n
o
n
ap
p
r
o
x
i
m
at
io
n
co
m
p
o
n
e
n
t
o
b
tain
ed
b
y
f
il
ter
in
g
co
-
ch
a
n
n
el
s
p
ee
c
h
at
o
n
e
d
is
cr
ete
w
a
v
elet
tr
an
s
f
o
r
m
(
DW
T
)
s
ca
le.
T
h
ese
m
et
h
o
d
s
s
h
o
w
t
h
at
t
h
e
s
p
ea
k
er
id
en
ti
f
icatio
n
s
y
s
te
m
ac
h
i
ev
es
ap
p
r
o
x
i
m
a
tel
y
8
0
% o
f
co
r
r
ec
t id
en
tif
icatio
n
w
h
e
n
th
e
o
v
er
all
T
I
R
is
2
0
d
B
.
I
n
o
u
r
p
r
ev
io
u
s
w
o
r
k
[
1
0
-
11
]
,
w
e
h
av
e
d
e
v
elo
p
ed
m
u
lt
i
r
eso
lu
tio
n
d
y
ad
ic
w
a
v
elet
(
MRDW
T
)
m
et
h
o
d
to
d
etec
t
u
s
ab
le
s
p
ee
ch
.
MR
DW
T
m
et
h
o
d
is
a
lin
ea
r
m
u
lti
-
s
ca
le
d
ec
o
m
p
o
s
i
tio
n
w
h
ich
ap
p
lie
s
d
is
c
r
ete
w
a
v
elet
tr
an
s
f
o
r
m
(
D
W
T
)
iter
ativ
el
y
to
d
etec
t
p
itch
p
er
io
d
icity
.
W
e
ar
e
m
o
ti
v
ated
b
y
d
etec
ti
n
g
p
itc
h
in
f
o
r
m
atio
n
i
n
all
lo
w
er
f
r
eq
u
en
c
y
s
u
b
-
b
an
d
s
o
f
co
-
ch
a
n
n
el
s
p
ee
ch
.
I
n
f
ac
t,
u
s
ab
le
f
r
a
m
e
s
ar
e
ch
ar
ac
ter
ized
b
y
p
er
io
d
icit
y
f
ea
t
u
r
es.
T
h
e
MRDW
T
m
et
h
o
d
g
iv
es
g
o
o
d
h
its
p
er
ce
n
ta
g
e
.
T
h
e
d
etec
ted
u
s
ab
le
s
e
g
m
e
n
t
s
ar
e
s
ep
ar
ated
in
ti
m
e
an
d
n
ee
d
to
b
e
o
r
g
an
ized
in
to
s
p
ea
k
er
s
tr
ea
m
s
.
R
ec
en
tl
y
,
w
e
h
a
v
e
p
r
o
p
o
s
ed
in
[
12
]
a
sp
ea
k
er
as
s
ig
n
m
e
n
t
s
y
s
te
m
th
at
o
r
g
an
ize
s
u
s
ab
le
s
p
ee
c
h
s
e
g
m
en
ts
u
n
d
er
co
-
ch
a
n
n
el
co
n
d
i
tio
n
s
.
T
h
en
,
u
s
ab
le
s
eg
m
e
n
ts
ar
e
a
s
s
i
g
n
ed
to
t
wo
s
p
ea
k
er
g
r
o
u
p
s
,
co
r
r
esp
o
n
d
in
g
to
th
e
t
w
o
s
p
ea
k
er
s
i
n
t
h
e
m
i
x
t
u
r
e.
Fin
a
ll
y
,
s
p
ea
k
er
s
ar
e
id
en
ti
f
ied
u
s
in
g
t
h
e
ass
ig
n
ed
s
eg
m
e
n
t
s
.
I
n
th
i
s
p
ap
er
,
w
e
p
r
o
p
o
s
e
to
e
v
alu
a
te
co
-
c
h
an
n
el
s
p
ea
k
er
id
en
ti
f
icatio
n
s
y
s
te
m
.
MR
DW
T
is
u
s
ed
f
o
r
co
-
ch
a
n
n
e
l
s
p
ee
ch
p
r
o
ce
s
s
i
n
g
.
E
v
al
u
atio
n
is
p
er
f
o
r
m
ed
o
n
T
I
MI
T
d
atab
ase.
T
h
is
p
ap
er
is
o
r
g
an
ized
as
f
o
llo
w
s
.
I
n
s
ec
tio
n
2
,
th
e
li
n
ea
r
m
u
lti
-
s
ca
le
m
et
h
o
d
is
p
r
esen
ted
.
I
n
s
ec
tio
n
3
,
th
e
co
-
ch
a
n
n
e
l
s
p
ea
k
er
id
en
ti
f
icatio
n
s
y
s
te
m
i
s
p
r
esen
ted
.
T
h
e
ex
p
er
im
e
n
tal
r
es
u
lt
s
o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
i
n
th
i
s
p
ap
er
an
d
th
e
tar
g
et
s
p
ea
k
er
id
en
t
if
icatio
n
ac
cu
r
ac
y
ar
e
also
p
r
esen
ted
i
n
s
ec
tio
n
4
.
Fin
all
y
,
o
u
r
w
o
r
k
o
f
t
h
is
p
ap
er
is
s
u
m
m
ar
ized
in
t
h
e
last
s
ec
tio
n
.
2.
L
I
N
E
AR
M
UL
T
-
SCA
L
E
D
E
CO
M
P
O
SI
T
I
O
N
F
O
R
US
AB
L
E
SP
E
E
CH
D
E
T
E
C
T
I
O
N
Usab
le
f
r
a
m
es
ar
e
c
h
ar
ac
ter
ized
b
y
p
er
io
d
icit
y
f
ea
t
u
r
es.
T
h
ese
f
ea
t
u
r
es
s
h
o
u
ld
b
e
lo
ca
t
ed
in
lo
w
-
f
r
eq
u
en
c
y
b
a
n
d
th
a
t
i
n
clu
d
es
th
e
p
itc
h
f
r
eq
u
e
n
c
y
.
L
i
n
ea
r
Mu
lti
-
s
ca
le
d
ec
o
m
p
o
s
itio
n
b
ased
o
n
DW
T
is
ap
p
lied
iter
ativ
el
y
in
o
r
d
er
to
d
eter
m
in
e
t
h
e
s
u
i
tab
le
b
an
d
f
o
r
p
e
r
io
d
icity
d
etec
tio
n
.
I
n
th
i
s
b
an
d
,
p
er
io
d
icity
f
ea
t
u
r
es a
r
e
n
o
t
m
u
c
h
d
is
t
u
r
b
ed
b
y
i
n
ter
f
er
er
s
p
ee
ch
i
n
ca
s
e
o
f
u
s
ab
le
s
e
g
m
en
t
s
.
I
n
ca
s
e
o
f
u
n
u
s
ab
le
f
r
a
m
es,
i
t
is
n
o
t
p
o
s
s
ib
le
to
d
etec
t
p
er
io
d
ici
t
y
i
n
all
lo
w
er
s
u
b
-
b
an
d
s
.
A
t
ea
ch
iter
atio
n
,
a
u
to
co
r
r
elatio
n
i
s
ap
p
lied
to
t
h
e
ap
p
r
o
x
im
a
tio
n
co
ef
f
icie
n
t
s
in
o
r
d
er
t
o
d
etec
t
p
e
r
io
d
icity
[
1
1
]
.
T
h
r
ee
d
o
m
in
a
ted
lo
ca
l
m
ax
i
m
a
ar
e
d
eter
m
in
ed
f
r
o
m
t
h
e
a
u
to
co
r
r
elatio
n
s
i
g
n
al
w
ith
a
p
ea
k
-
p
ick
in
g
al
g
o
r
it
h
m
w
h
ich
u
s
es
a
t
h
r
es
h
o
ld
ca
lcu
lated
f
r
o
m
lo
ca
l
m
ax
i
m
a
a
m
p
lit
u
d
es.
A
d
if
f
er
en
ce
o
f
au
to
co
r
r
elatio
n
lag
b
et
w
ee
n
th
e
f
ir
s
t
a
n
d
s
ec
o
n
d
m
a
x
i
m
u
m
a
n
d
b
et
w
ee
n
th
e
s
ec
o
n
d
a
n
d
th
ir
d
m
a
x
i
m
u
m
is
d
eter
m
i
n
ed
.
I
f
t
h
is
d
i
f
f
er
en
ce
i
s
les
s
th
a
n
t
h
e
th
r
e
s
h
o
ld
,
p
er
io
d
icity
is
d
etec
ted
an
d
co
-
c
h
an
n
el
s
p
ee
c
h
s
e
g
m
e
n
t
i
s
cla
s
s
i
f
ied
a
s
u
s
a
b
le.
T
h
is
th
r
e
s
h
o
ld
i
s
e
m
p
ir
ica
ll
y
f
i
x
ed
ac
co
r
d
in
g
to
th
e
b
est
ev
a
lu
at
io
n
r
es
u
lt
s
.
T
h
e
o
p
tim
u
m
t
h
r
es
h
o
ld
v
al
u
e
o
f
8
s
a
m
p
les
is
c
h
o
s
e
n
at
1
6
k
Hz
s
a
m
p
li
n
g
f
r
eq
u
en
c
y
.
I
f
at
t
h
i
s
iter
atio
n
,
p
er
io
d
icity
is
n
o
t
d
etec
ted
,
DW
T
is
ap
p
lied
to
a
p
p
r
o
x
im
ati
o
n
s
ig
n
al
i
n
o
r
d
er
to
d
etec
t
h
id
d
en
p
er
io
d
icit
y
f
ea
t
u
r
e
i
n
f
in
er
b
a
n
d
f
r
eq
u
en
c
y
.
Fo
r
u
n
u
s
ab
le
f
r
a
m
es,
it
i
s
n
o
t
p
o
s
s
ib
le
to
d
etec
t
p
er
io
d
icity
i
n
all
lo
w
er
s
u
b
-
b
a
n
d
s
.
A
m
a
x
i
m
u
m
o
f
4
iter
atio
n
s
ar
e
allo
w
ed
.
T
h
is
li
m
it
is
f
i
x
ed
b
ased
o
n
p
itch
b
an
d
.
T
h
e
lo
w
e
s
t b
an
d
s
h
o
u
ld
co
r
r
esp
o
n
d
to
p
itch
b
an
d
.
Fig
u
r
e
1
co
r
r
esp
o
n
d
s
to
a
u
s
ab
le
f
r
a
m
e
f
o
r
m
ale
-
m
a
le
co
-
c
h
an
n
el
s
p
ee
c
h
.
I
n
t
h
is
ca
s
e,
p
er
io
d
icity
i
s
d
etec
ted
o
n
l
y
at
s
ca
le
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
6
,
No
.
6
,
Decem
b
er
201
6
: 2
7
6
6
–
2
7
7
2
2768
Fig
u
r
e
1
.
An
al
y
s
is
o
f
a
Usab
le
Sp
ee
ch
Fra
m
e
f
o
r
Ma
le
-
Ma
le
Co
-
C
h
a
n
n
e
l
Sp
ee
ch
U
p
to
Sc
a
le
3
3.
CO
-
CH
ANNE
L
SPEAK
E
R
I
DE
N
T
I
F
I
CA
T
I
O
N
S
YST
E
M
I
n
o
r
d
er
to
id
en
tify
t
h
e
tar
g
et
an
d
th
e
i
n
ter
f
er
er
s
p
ea
k
er
s
,
th
e
d
etec
ted
u
s
ab
le
s
eg
m
e
n
ts
ar
e
o
r
g
an
ized
in
to
t
w
o
s
p
ea
k
er
s
tr
ea
m
s
b
y
a
s
p
ea
k
er
ass
i
g
n
m
en
t
s
y
s
te
m
[
1
2
]
.
T
h
e
s
p
ea
k
er
ass
ig
n
m
en
t
s
y
s
te
m
o
r
g
an
ize
s
u
s
ab
le
s
p
ee
ch
s
e
g
m
e
n
ts
u
n
d
er
co
-
ch
an
n
el
co
n
d
itio
n
s
.
I
t
h
as
ex
ten
d
ed
p
r
o
b
ab
ilis
tic
f
r
a
m
e
w
o
r
k
o
f
tr
ad
itio
n
a
l
SID
to
co
-
ch
an
n
el
s
p
ee
ch
.
I
t
u
s
es
ex
h
a
u
s
ti
v
e
s
ea
r
ch
al
g
o
r
ith
m
to
m
a
x
i
m
ize
t
h
e
p
o
s
ter
io
r
p
r
o
b
a
b
ilit
y
in
g
r
o
u
p
in
g
u
s
ab
le
s
p
ee
ch
.
T
h
en
,
u
s
ab
le
s
e
g
m
en
t
s
ar
e
a
s
s
i
g
n
ed
to
t
w
o
s
p
ea
k
er
g
r
o
u
p
s
,
co
r
r
es
p
o
n
d
in
g
to
t
h
e
t
w
o
s
p
ea
k
er
s
i
n
t
h
e
m
i
x
t
u
r
e.
T
h
e
t
w
o
s
p
ea
k
er
s
tr
ea
m
s
ar
e
u
s
e
d
as
in
p
u
t
f
o
r
a
b
aseli
n
e
s
p
e
ak
er
id
en
ti
f
icat
io
n
s
y
s
te
m
.
3
.
1
.
Sp
ea
k
er
As
s
ig
ne
m
e
nt
I
n
s
p
ea
k
er
id
en
ti
f
icat
io
n
s
y
s
te
m
,
d
is
cr
i
m
in
at
io
n
b
et
w
ee
n
s
p
ea
k
er
s
is
b
ased
o
n
p
o
s
ter
io
r
p
r
o
b
ab
ilit
y
.
T
h
e
g
o
al
i
s
to
f
in
d
t
h
e
s
p
ea
k
er
m
o
d
el
r
ef
er
en
ce
i
n
th
e
s
et
o
f
s
p
ea
k
er
m
o
d
els
Λ={
λ
1
,
λ
2
,
…,
λ
k
},
th
a
t
m
ax
i
m
izes
t
h
e
p
o
s
ter
io
r
p
r
o
b
ab
ilit
y
f
o
r
an
o
b
s
er
v
at
i
o
n
s
e
q
u
en
ce
O=
{o
1
,
o
2
,
…,
o
M}
[
2
]
.
C
ep
s
tr
al
f
ea
t
u
r
es,
s
u
c
h
as
m
el
-
f
r
eq
u
en
c
y
ce
p
s
tr
al
co
ef
f
icie
n
ts
(
MF
C
C
s
)
,
ar
e
u
s
ed
as
o
b
s
er
v
atio
n
s
f
o
r
s
p
ee
ch
s
ig
n
al
s
.
T
h
e
g
o
al
in
co
-
c
h
a
n
n
el
at
te
m
p
t
s
to
f
in
d
t
w
o
s
p
ea
k
er
m
o
d
el
s
th
at
m
a
x
i
m
ize
t
h
e
p
o
s
ter
io
r
p
r
o
b
a
b
ilit
y
f
o
r
t
h
e
o
b
s
er
v
atio
n
s
.
I
n
[
1
2
]
,
w
e
h
a
v
e
p
r
o
p
o
s
ed
a
s
p
ea
k
er
ass
i
g
n
m
en
t
s
y
s
te
m
th
at
o
r
g
a
n
izes
u
s
ab
le
s
p
ee
ch
s
eg
m
e
n
ts
u
n
d
er
co
-
c
h
an
n
el
co
n
d
itio
n
s
.
W
e
h
av
e
ex
te
n
d
ed
p
r
o
b
ab
ili
s
tic
f
r
a
m
e
w
o
r
k
o
f
tr
ad
itio
n
al
SID
to
co
-
ch
an
n
el
s
p
ee
ch
.
Fo
r
a
co
-
ch
an
n
el
m
i
x
tu
r
e,
o
u
r
u
s
ab
le
s
p
ee
ch
ex
t
r
ac
tio
n
m
et
h
o
d
ex
tr
ac
ts
N
c
o
n
s
ec
u
tiv
e
s
p
ee
ch
s
eg
m
e
n
ts
,
X=
{S
1
,S
2
,
…,
S
i
,
…,
S
N
}.
Usab
le
s
e
g
m
en
ts
ar
e
o
r
g
an
ize
d
in
to
t
w
o
s
p
ea
k
er
s
tr
ea
m
s
b
ec
au
s
e
in
co
-
c
h
an
n
el
s
p
ee
ch
o
n
e
s
p
ea
k
er
ca
n
d
o
m
i
n
ate
in
s
o
m
e
p
o
r
tio
n
s
a
n
d
b
e
d
o
m
i
n
ated
i
n
o
t
h
er
p
o
r
tio
n
s
.
Fo
r
e
x
a
m
p
le,
a
p
o
s
s
ib
le
s
e
g
m
en
t
ass
i
g
n
m
e
n
t
m
a
y
lo
o
k
li
k
e
0
1
0
1
1
2
i
N
S
,
S
,
,
S
,
,
S
,
w
h
er
e
s
u
p
er
s
cr
ip
ts
,
0
an
d
1
,
d
o
n
o
t
r
e
p
r
esen
t
th
e
s
p
ea
k
e
r
id
en
titi
e
s
b
u
t
o
n
l
y
i
n
d
icate
t
h
a
t
th
e
s
eg
m
e
n
t
s
m
ar
k
ed
w
it
h
t
h
e
s
a
m
e
lab
el
ar
e
f
r
o
m
th
e
s
a
m
e
s
p
ea
k
er
.
I
n
[
1
2
]
,
w
e
h
av
e
d
e
m
o
n
s
tr
ated
th
at
p
r
o
b
ab
ilit
y
p
o
s
ter
io
r
ca
n
b
e
r
e
-
w
r
itten
f
o
r
co
-
ch
a
n
n
e
l sp
ee
ch
as
:
01
01
P
X
,
|
λ
,
λ
P
S
|
λ
P
S
|
λ
j
i
I
I
I
i
I
j
I
I
S
X
S
X
X
T
h
e
p
r
o
b
a
b
ilit
y
o
f
h
a
v
i
n
g
a
s
eg
m
e
n
t
S
f
r
o
m
a
p
r
e
-
tr
ain
ed
s
p
ea
k
er
m
o
d
el
λ
i
s
t
h
e
p
r
o
d
u
ct
o
f
lik
eli
h
o
o
d
s
o
f
th
at
s
p
ea
k
er
m
o
d
el
g
en
er
ati
n
g
ea
c
h
in
d
i
v
id
u
a
l
o
b
s
er
v
atio
n
x
o
f
t
h
e
s
e
g
m
en
t
.
Sirig
o
s
e
t
al
[
1
3
]
an
d
L
o
v
e
k
i
n
et
al
[
1
]
h
av
e
s
h
o
w
n
th
at
v
o
iced
s
p
ee
ch
p
la
y
s
a
d
o
m
i
n
an
t
r
o
le
i
n
s
p
ea
k
er
r
ec
o
g
n
itio
n
.
T
h
e
id
ea
o
f
u
s
i
n
g
o
n
l
y
t
h
e
v
o
iced
p
ar
t
o
f
s
p
ee
ch
s
i
g
n
al
i
s
b
ased
o
n
th
e
f
ac
t
t
h
at
v
o
iced
s
p
ee
ch
s
eg
m
e
n
t
co
n
tai
n
th
e
m
o
s
t
s
i
g
n
i
f
ican
t
s
p
ea
k
er
id
en
ti
f
icatio
n
as
o
p
p
o
s
ed
t
o
o
th
er
s
p
ee
ch
s
eg
m
en
t.
0
50
100
150
200
250
300
350
400
450
500
-
0
.
5
0
0
.
5
C
o
-
c
h
a
n
n
e
l
s
p
e
e
c
h
0
50
100
150
200
250
300
350
400
450
500
-1
0
1
D
W
T
a
p
p
r
o
x
i
m
a
t
i
o
n
a
t
s
c
a
l
e
1
0
50
100
150
200
250
300
350
400
-
0
.
5
0
0
.
5
D
W
T
a
p
p
r
o
x
i
m
a
t
i
o
n
a
t
s
c
a
l
e
3
0
100
200
300
400
500
600
-5
0
5
A
u
t
o
c
o
r
r
e
l
a
t
i
o
n
o
f
a
p
p
r
o
x
i
m
a
t
i
o
n
a
t
s
c
a
l
e
3
l
a
g
0
50
100
150
200
250
300
350
400
450
500
-
0
.
5
0
0
.
5
D
W
T
a
p
p
r
o
x
i
m
a
t
i
o
n
a
t
s
c
a
l
e
2
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
On
Usa
b
le
S
p
ee
ch
Dete
ctio
n
b
y
Lin
ea
r
Mu
lti
-
S
ca
le
Dec
o
mp
o
s
itio
n
fo
r
S
I
D
(
W
a
jd
i G
h
eza
i
el
)
2769
W
h
en
v
o
iced
o
n
l
y
s
e
g
m
e
n
ts
w
er
e
u
s
ed
f
o
r
tr
ain
in
g
an
d
tes
tin
g
ap
p
r
o
x
i
m
atel
y
8
0
%
s
p
ea
k
er
id
en
tit
y
ac
c
u
r
ac
y
w
a
s
ac
h
ie
v
ed
.
T
h
er
ef
o
r
e,
w
e
p
r
o
p
o
s
e
to
u
s
e
v
o
iced
f
r
a
m
e
i
n
tr
ain
i
n
g
.
Ob
s
er
v
atio
n
s
ar
e
ex
t
r
ac
ted
f
r
o
m
v
o
ic
ed
f
r
a
m
e
b
y
MF
C
C
s
.
Sp
ea
k
er
s
m
o
d
el
s
ar
e
f
o
r
m
ed
w
it
h
1
6
-
m
i
x
tu
r
e
GM
M
s
.
W
e
e
m
p
lo
y
ex
h
au
s
ti
v
e
s
ea
r
c
h
alg
o
r
ith
m
to
f
i
n
d
co
r
r
esp
o
n
d
en
t
s
p
ea
k
er
.
I
n
i
m
p
le
m
en
ta
tio
n
,
th
e
r
ea
l
co
m
p
u
tatio
n
ti
m
e
i
s
lo
n
g
er
.
I
t
ca
n
b
e
f
u
r
t
h
er
r
ed
u
ce
d
b
y
s
to
r
in
g
all
th
e
li
k
eli
h
o
o
d
s
co
r
es
o
f
a
s
eg
m
en
t
g
i
v
e
n
a
m
o
d
el
in
t
h
e
m
e
m
o
r
y
as
a
tab
le
an
d
lo
o
k
in
g
u
p
a
s
co
r
e
f
r
o
m
t
h
e
ta
b
le
w
h
en
n
ee
d
ed
.
3
.
2
.
B
a
s
eline
SI
D
Sy
s
t
e
m
T
h
e
SID
is
p
er
f
o
r
m
ed
w
ith
a
b
aselin
e
s
y
s
te
m
[
2
-
3]
.
Mo
d
elin
g
is
a
s
s
u
r
ed
b
y
Ga
u
s
s
ia
n
Mi
x
t
u
r
e
Mo
d
el
(
GM
M)
an
d
esti
m
ated
t
h
r
o
u
g
h
t
h
e
E
x
p
ec
tatio
n
Ma
x
i
m
izati
o
n
(
E
M)
alg
o
r
ith
m
t
h
at
m
a
x
i
m
izes
th
e
lik
e
lih
o
o
d
cr
iter
io
n
.
A
s
et
o
f
1
6
m
ix
t
u
r
es
ar
e
u
s
ed
f
o
r
s
p
ea
k
er
m
o
d
el.
I
n
o
u
r
ex
p
er
i
m
e
n
t,
w
e
u
s
e
th
e
clas
s
ical
p
ar
am
eter
izat
io
n
b
ased
o
n
1
6
Me
l
Fre
q
u
e
n
c
y
C
ep
s
tr
al
C
o
ef
f
ic
ien
t
s
(
MF
C
C
)
.
T
h
ese
co
ef
f
ic
ie
n
t
s
ar
e
co
m
p
u
ted
f
r
o
m
th
e
s
p
ee
c
h
s
ig
n
a
l
ev
er
y
1
0
m
s
u
s
in
g
a
ti
m
e
w
in
d
o
w
o
f
2
5
m
s
.
E
ac
h
f
ea
t
u
r
e
v
ec
to
r
is
p
r
esen
ted
b
y
t
h
e
m
id
d
le
w
in
d
o
w
s
o
f
e
v
er
y
u
t
ter
an
ce
.
Sp
ea
k
er
m
o
d
el
is
tr
ain
ed
u
s
i
n
g
th
e
E
M
alg
o
r
ith
m
w
it
h
th
e
f
ea
t
u
r
es
ca
lc
u
lated
f
r
o
m
t
r
ain
in
g
s
a
m
p
l
e
s
.
I
n
test
in
g
p
h
ase,
t
h
e
o
r
g
a
n
ized
u
s
ab
le
s
p
ee
ch
,
w
it
h
s
p
ea
k
er
ass
i
g
n
m
e
n
t
s
y
s
te
m
,
ar
e
u
s
ed
as
test
s
p
ee
ch
s
a
m
p
les
f
o
r
SID
s
y
s
te
m
.
T
h
e
s
a
m
e
f
ea
tu
r
es
ar
e
d
er
iv
ed
f
r
o
m
t
h
e
test
s
p
ee
ch
s
a
m
p
les
a
n
d
ar
e
in
p
u
t
to
ev
er
y
s
p
ea
k
er
‟
s
GM
M.
T
h
e
s
p
ea
k
er
w
i
th
t
h
e
h
i
g
h
est
li
k
eli
h
o
o
d
s
co
r
e
r
ep
r
esen
ts
th
e
id
e
n
ti
f
ied
s
p
ea
k
er
.
Her
e,
s
p
ea
k
er
id
en
tif
icatio
n
ex
p
er
i
m
en
t
s
ar
e
clo
s
e
-
s
et
an
d
tex
t
-
i
n
d
ep
en
d
en
t.
4.
E
XP
E
R
I
M
E
NT
AND
R
E
SU
L
T
Sp
ee
ch
d
ata
f
r
o
m
t
h
e
T
I
MI
T
d
atab
ase
w
a
s
u
s
ed
f
o
r
all
t
h
e
s
i
m
u
lat
io
n
e
x
p
er
i
m
e
n
ts
.
T
h
e
s
p
ea
k
er
s
et
is
co
m
p
o
s
ed
o
f
3
8
s
p
ea
k
er
s
f
r
o
m
t
h
e
“
D
R
1
”
d
ialec
t
r
eg
io
n
,
1
4
o
f
w
h
ich
ar
e
f
e
m
ale
an
d
th
e
r
est
ar
e
m
ale.
Fo
r
ea
ch
p
air
o
f
s
p
ea
k
er
th
e
T
I
R
is
ca
lcu
lated
as
th
e
en
er
g
y
r
atio
o
f
th
e
tar
g
et
s
p
ee
ch
o
v
er
th
e
in
ter
f
er
e
n
ce
s
p
ee
ch
.
T
h
r
ee
d
if
f
er
en
t
s
et
s
o
f
co
-
ch
a
n
n
e
l
s
p
ee
ch
ar
e
co
n
s
id
er
ed
:
m
ale
-
m
ale,
f
e
m
ale
-
f
e
m
ale,
an
d
m
ale
-
f
e
m
ale.
Sp
ee
ch
s
ig
n
al
s
ar
e
s
ca
led
to
c
r
ea
te
th
e
m
i
x
t
u
r
es
at
d
if
f
er
en
t
T
I
R
s
:
-
2
0
d
B
,
-
1
0
d
B
,
-
5
d
B
,
0
d
B
,
5
d
B
,
1
0
d
B
an
d
2
0
d
B
.
4
.
1
.
M
RDWT E
v
a
lua
t
io
n
T
h
e
T
ar
g
et
to
I
n
ter
f
er
er
R
atio
T
I
R
m
ea
s
u
r
e
i
s
u
s
ed
to
lab
el
v
o
iced
f
r
a
m
es
a
s
u
s
ab
le
o
r
u
n
u
s
ab
le.
Fo
r
u
s
ab
ilit
y
d
ec
is
io
n
,
f
r
a
m
e
s
th
a
t
h
av
e
ab
o
v
e
2
0
d
B
T
I
R
ar
e
c
o
n
s
id
er
ed
as
u
s
ab
le.
E
v
al
u
ati
o
n
is
b
ased
o
n
h
it
s
an
d
f
alse a
lar
m
s
p
er
ce
n
tag
e
s
.
T
h
e
p
er
f
o
r
m
a
n
ce
o
f
p
r
o
p
o
s
ed
m
et
h
o
d
is
g
iv
e
n
i
n
T
ab
le
1
.
W
e
co
m
p
ar
e
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
w
ith
r
elate
d
ap
p
r
o
ac
h
es
in
[
9
]
.
On
av
er
ag
e
t
h
e
MRDW
T
m
et
h
o
d
d
etec
ts
at
least
9
5
.
7
6
%
o
f
th
e
u
s
ab
le
s
p
ee
ch
w
it
h
a
f
alse
alar
m
r
ate
o
f
2
9
.
6
5
%.
P
ea
k
d
if
f
er
en
ce
au
to
co
r
r
elatio
n
o
f
w
a
v
elet
tr
an
s
f
o
r
m
(
P
DA
W
T
)
m
et
h
o
d
[
9
]
is
b
ased
o
n
p
itch
i
n
f
o
r
m
atio
n
d
etec
tio
n
.
T
h
is
m
et
h
o
d
ap
p
lies
DW
T
o
n
ce
o
n
l
y
to
co
-
c
h
a
n
n
el
s
p
ee
ch
to
d
etec
t
p
itch
in
f
o
r
m
atio
n
.
On
a
v
er
ag
e
th
e
P
DA
W
T
m
et
h
o
d
d
etec
ts
at
least
8
0
%
o
f
th
e
u
s
ab
le
s
p
ee
ch
w
it
h
a
f
alse
alar
m
r
ate
o
f
3
0
%.
T
ab
le
1
.
R
esu
lts
o
f
P
DA
W
T
an
d
MRDW
T
Met
h
o
d
f
o
r
U
s
ab
le
Sp
ee
ch
Dete
ct
io
n
Co
-
c
h
a
n
n
e
l
P
D
A
W
T
M
R
D
W
T
%
H
i
t
%
F
a
l
s
e
a
l
a
r
m
%
H
i
t
%
F
a
l
s
e
a
l
a
r
m
F
e
mal
e
-
F
e
mal
e
8
2
.
0
3
2
.
3
9
3
.
0
2
3
2
.
3
7
M
a
l
e
-
M
a
l
e
8
0
.
5
3
0
.
6
9
8
.
4
6
2
8
.
9
3
M
a
l
e
-
F
e
mal
e
8
1
.
3
2
9
.
6
9
5
.
8
0
2
7
.
6
6
A
v
e
r
a
g
e
8
1
.
2
3
0
.
8
9
5
.
7
6
2
9
.
6
5
MRDW
T
ac
h
iev
e
a
m
ax
i
m
u
m
o
f
d
etec
ted
u
s
ab
le
s
p
ee
c
h
co
m
p
ar
ed
to
P
DA
W
T
.
W
e
c
o
n
s
id
er
th
e
ef
f
ec
tiv
e
n
e
s
s
o
f
li
n
ea
r
m
u
lt
i
-
s
ca
le
d
ec
o
m
p
o
s
itio
n
b
y
MR
D
W
T
to
in
cr
ea
s
e
th
e
p
er
ce
n
t o
f
h
it.
4
.
2
.
Sp
ea
k
er
I
dentif
ica
t
io
n E
v
a
lua
t
io
n
T
o
d
em
o
n
s
tr
ate
t
h
e
u
s
e
f
u
l
n
e
s
s
o
f
o
u
r
p
r
o
p
o
s
ed
m
e
th
o
d
,
u
s
ab
le
s
p
ee
ch
is
a
s
s
i
g
n
ed
in
to
s
tr
ea
m
s
b
y
o
u
r
s
p
ea
k
er
ass
ig
n
m
e
n
t
s
y
s
te
m
.
I
f
th
e
tar
g
et
s
p
ea
k
er
is
o
f
in
ter
est,
th
e
n
th
e
s
p
ee
ch
s
i
g
n
al
f
r
o
m
t
h
e
o
th
er
s
p
ea
k
er
is
co
n
s
id
er
ed
n
o
is
e.
W
e
ch
o
o
s
e
th
e
tar
g
et
s
p
ea
k
er
SID
as
o
u
r
e
v
alu
a
tio
n
cr
iter
i
o
n
.
Fi
g
u
r
e
2
s
h
o
ws
co
m
b
i
n
atio
n
o
f
co
-
c
h
an
n
el
f
e
m
ale
-
m
ale.
W
e
s
h
o
w
th
e
co
r
r
esp
o
n
d
en
t a
s
s
ig
n
m
e
n
t to
s
p
ea
k
er
1
an
d
2
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
6
,
No
.
6
,
Decem
b
er
201
6
: 2
7
6
6
–
2
7
7
2
2770
Fig
u
r
e
2
.
Sp
ea
k
er
Ass
i
g
n
m
e
n
t
f
o
r
Fe
m
ale
-
Ma
le
C
o
-
C
h
a
n
n
el
Fig
u
r
e
3
.
P
er
f
o
r
m
a
n
ce
o
f
th
e
P
r
o
p
o
s
ed
Sp
ea
k
er
I
d
en
tif
icati
o
n
u
n
d
er
Co
-
C
h
a
n
n
e
l Co
n
d
iti
o
n
s
C
o
m
p
ar
ed
w
it
h
Re
lated
Me
th
o
d
s
Fig
u
r
e
3
g
i
v
es
t
h
e
tar
g
e
t
SID
co
r
r
ec
t
ac
cu
r
ac
y
f
o
r
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
an
d
r
elate
d
m
eth
o
d
.
I
n
[
1
4
]
Sh
ao
p
r
o
p
o
s
ed
r
o
b
u
s
t p
itc
h
tr
a
ck
in
g
m
e
th
o
d
to
e
x
tr
ac
t
u
s
ab
l
e
s
p
ee
ch
f
o
r
s
p
ea
k
er
id
en
ti
f
ica
tio
n
ta
s
k
.
B
ased
o
n
p
itch
i
n
f
o
r
m
atio
n
,
th
i
s
m
et
h
o
d
ex
tr
ac
ts
th
e
u
s
ab
le
s
p
ee
ch
s
eg
m
e
n
t
s
t
h
at
co
n
s
i
s
t
o
f
o
n
l
y
o
n
e
s
p
ea
k
er
‟
s
p
itc
h
an
d
f
ee
d
th
e
m
in
to
a
s
p
ea
k
er
id
en
tific
atio
n
s
y
s
te
m
.
T
ar
g
et
SID
co
r
r
ec
t
r
ate
ar
e
g
iv
en
b
ef
o
r
e
an
d
af
ter
u
s
ab
le
s
p
ee
ch
e
x
tr
ac
tio
n
.
I
t
‟
s
clea
r
f
r
o
m
Fi
g
u
r
e
3
th
at
t
h
e
M
R
D
W
T
p
e
r
f
o
r
m
s
s
ig
n
i
f
ica
n
tl
y
b
e
tter
th
a
n
M
u
lti
-
p
itc
h
tr
ac
k
in
g
an
d
P
DA
W
T
u
s
ab
le
s
p
ee
ch
m
et
h
o
d
s
.
MRDW
T
,
Mu
lti
-
p
itc
h
tr
ac
k
i
n
g
a
n
d
P
DA
W
T
u
s
ab
le
s
p
ee
ch
m
et
h
o
d
s
an
d
s
p
ea
k
er
as
s
ig
n
m
e
n
t
s
y
s
te
m
i
m
p
r
o
v
e
s
s
i
g
n
i
f
ican
tl
y
SID
p
er
f
o
r
m
an
ce
s
u
n
d
er
c
o
-
ch
a
n
n
el
s
itu
a
tio
n
s
.
T
h
e
ac
cu
r
ac
y
d
eg
r
ad
es
s
h
ar
p
l
y
w
h
e
n
T
I
R
d
ec
r
ea
s
es
b
ec
au
s
e
th
e
tar
g
e
t
s
p
ee
ch
is
in
cr
ea
s
in
g
l
y
co
r
r
u
p
ted
.
T
h
e
f
ir
s
t
o
b
s
er
v
ati
o
n
f
r
o
m
th
e
f
ig
u
r
e
is
th
at,
u
n
d
er
co
-
ch
an
n
el
s
it
u
atio
n
s
,
t
h
e
tar
g
et
SID
co
r
r
ec
t
r
ate
w
ith
u
s
ab
le
s
p
ee
c
h
d
etec
tio
n
i
s
b
etter
t
h
an
th
e
tar
g
et
SID
co
r
r
ec
t
r
ate
w
ith
o
u
t
u
s
ab
le
s
p
ee
ch
d
etec
t
io
n
.
T
h
e
p
r
o
p
o
s
ed
u
s
ab
le
s
p
ee
c
h
d
etec
tio
n
i
m
p
r
o
v
e
s
s
p
ea
k
er
id
en
ti
f
icatio
n
p
er
f
o
r
m
an
ce
.
Seco
n
d
l
y
,
th
e
i
m
p
r
o
v
e
m
en
t
s
ar
e
co
n
s
is
te
n
t
ac
r
o
s
s
all
T
I
R
lev
el
s
.
P
er
f
o
r
m
a
n
ce
i
m
p
r
o
v
e
m
en
t
in
cr
ea
s
e
s
at
h
i
g
h
er
T
I
R
s
b
ec
au
s
e
t
h
e
tar
g
e
t
s
p
ea
k
er
d
o
m
i
n
ates
th
e
m
i
x
t
u
r
e.
Ho
w
ev
e
r
,
tar
g
et
s
p
ea
k
er
i
s
d
o
m
i
n
ated
b
y
i
n
ter
f
er
e
n
ce
at
lo
w
er
T
I
R
s
,
r
esu
lti
n
g
in
lo
w
er
p
er
f
o
r
m
a
n
ce
af
ter
u
s
ab
le
s
p
ee
ch
ex
tr
ac
tio
n
.
5.
CO
NCLU
SI
O
N
I
n
th
i
s
p
ap
er
,
w
e
h
a
v
e
e
m
p
lo
y
ed
a
n
e
w
m
et
h
o
d
f
o
r
u
s
ab
le
s
p
ee
ch
ex
tr
ac
tio
n
to
i
m
p
r
o
v
e
s
p
ea
k
er
id
en
ti
f
icatio
n
u
n
d
er
co
-
ch
a
n
n
el
s
p
ee
ch
.
U
s
ab
le
s
p
ee
ch
is
e
x
tr
ac
ted
b
ased
o
n
p
itc
h
i
n
f
o
r
m
atio
n
o
b
tain
ed
b
y
lin
ea
r
m
u
lti
-
s
ca
le
d
ec
o
m
p
o
s
it
i
o
n
MRDW
T
.
Ou
r
u
s
ab
le
s
p
ee
ch
ex
tr
ac
tio
n
m
et
h
o
d
p
r
o
d
u
ce
s
s
eg
m
e
n
ts
u
s
e
f
u
l
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
On
Usa
b
le
S
p
ee
ch
Dete
ctio
n
b
y
Lin
ea
r
Mu
lti
-
S
ca
le
Dec
o
mp
o
s
itio
n
fo
r
S
I
D
(
W
a
jd
i G
h
eza
i
el
)
2771
f
o
r
co
-
ch
a
n
n
e
l
SID
ac
r
o
s
s
v
ar
io
u
s
T
I
R
co
n
d
itio
n
s
.
MR
D
W
T
ac
h
iev
es
a
g
o
o
d
p
er
ce
n
t
o
f
u
s
ab
le
s
p
ee
c
h
d
etec
tio
n
.
I
n
co
m
p
ar
is
o
n
w
it
h
P
DW
A
T
m
e
th
o
d
,
o
u
r
p
r
o
p
o
s
ed
m
e
th
o
d
ac
h
ie
v
es
a
g
o
o
d
p
er
ce
n
t
o
f
co
r
r
ec
t
d
etec
tio
n
.
W
e
co
n
s
id
er
th
e
ef
f
ec
tiv
e
n
e
s
s
o
f
m
u
lt
i
-
s
ca
le
d
ec
o
m
p
o
s
itio
n
to
ex
tr
ac
t
u
s
a
b
le
s
p
ee
ch
.
Usab
le
s
eg
m
e
n
ts
ar
e
a
s
s
i
g
n
ed
to
t
w
o
s
p
ea
k
er
g
r
o
u
p
s
,
co
r
r
esp
o
n
d
in
g
to
t
h
e
t
w
o
s
p
ea
k
er
s
i
n
t
h
e
m
i
x
tu
r
e.
Or
g
an
ized
u
s
ab
le
s
p
ee
ch
ar
e
u
s
ed
as
in
p
u
t
to
s
p
ea
k
er
id
en
ti
f
icatio
n
s
y
s
te
m
.
W
e
h
a
v
e
s
h
o
w
n
th
a
t
th
e
p
r
o
p
o
s
ed
u
s
ab
le
s
p
ee
ch
d
etec
tio
n
ac
h
ie
v
es
g
o
o
d
SID
p
er
f
o
r
m
a
n
ce
an
d
it
p
er
f
o
r
m
s
s
i
g
n
i
f
ican
tl
y
b
etter
th
an
w
it
h
o
u
t
u
s
ab
le
s
p
ee
ch
d
etec
tio
n
.
SID
p
er
f
o
r
m
an
ce
d
eg
r
ad
es
w
h
en
T
I
R
d
ec
r
ea
s
es
b
ec
au
s
e
th
e
tar
g
et
s
p
ee
ch
is
in
cr
ea
s
in
g
l
y
co
r
r
u
p
ted
b
y
in
ter
f
er
er
s
p
ee
ch
.
RE
F
E
R
E
NC
E
S
[1
]
J.
L
o
v
e
k
in
,
R.
E.
Ya
n
to
r
n
o
,
S
.
Be
n
in
c
a
sa
,
S
.
W
e
n
n
d
t,
M
.
Hu
g
g
in
s,
De
v
e
lo
p
in
g
u
s
a
b
le
sp
e
e
c
h
c
riter
ia
fo
r
sp
e
a
k
e
r
id
e
n
ti
f
ica
ti
o
n
,
P
ro
c
.
ICA
S
S
P
2
0
0
1
–
P
.
4
2
1
-
4
2
4
.
[2
]
D.A
.
Re
y
n
o
ld
s.
Au
to
m
a
ti
c
sp
e
a
k
e
r re
c
o
g
n
it
i
o
n
u
sin
g
Gu
a
ss
i
a
n
mix
tu
re
sp
e
a
k
e
r mo
d
e
l
,
L
in
c
o
l
n
L
a
b
.
J.,
1
9
95
–
V
o
l
.
8
–
P
.
1
7
3
-
1
9
2
.
[3
]
Ka
y
o
d
e
F
ra
n
c
is
Ak
in
g
b
a
d
e
,
Ok
o
k
o
M
k
p
o
u
to
Um
a
n
n
a
,
Isia
k
a
A
je
w
a
l
e
A
li
m
i.
”
V
o
ice
-
Ba
se
d
Do
o
r
A
c
c
e
ss
Co
n
tro
l
S
y
st
e
m
Us
in
g
th
e
M
e
l
F
re
q
u
e
n
c
y
C
e
p
stru
m
Co
e
ff
icie
n
ts
a
n
d
G
a
u
ss
ian
M
ix
tu
re
M
o
d
e
l”.
I
n
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
t
e
r E
n
g
in
e
e
rin
g
(
IJ
ECE
),
V
o
l
4
N
o
5
,
2
0
1
4
.
[4
]
Di
W
u
,
Ch
i
n
a
;
Jie
Ca
o
;
H
u
a
Jin
W
a
n
g
,
“
S
p
e
a
k
e
r
Re
c
o
g
n
it
io
n
Ba
se
d
o
n
iv
e
c
to
r
a
n
d
Im
p
ro
v
e
d
L
o
c
a
l
P
re
se
rv
in
g
P
r
o
jec
ti
o
n
”
.
T
E
L
KOM
NIKA
In
d
o
n
e
sia
n
J
o
u
rn
a
l
o
f
E
lec
trica
l
En
g
i
n
e
e
rin
g
,
Vo
l
1
2
N
o
6
,
2
0
1
4
p
a
g
e
s
4
2
9
9
-
4
3
0
5
.
[5
]
R.
E
.
Ya
n
t
o
rn
o
,
“
Co
-
c
h
a
n
n
e
l
sp
e
e
c
h
stu
d
y
,
fi
n
a
l
re
p
o
rt
f
o
r
su
mm
e
r
re
se
a
rc
h
fa
c
u
lt
y
p
ro
g
ra
m
”
,
T
e
c
h
.
Re
p
.
,
A
ir
F
o
rc
e
Of
fi
c
e
o
f
S
c
ien
ti
fi
c
Re
se
a
rc
h
,
S
p
e
e
c
h
P
ro
c
e
ss
in
g
L
a
b
,
Ro
m
e
L
a
b
s,
Ne
w
Yo
rk
,
1
9
9
9
.
[6
]
Ya
n
to
rn
o
,
R
.
E,
M
e
th
o
d
f
o
r
im
p
ro
v
in
g
sp
e
a
k
e
r
id
e
n
ti
f
ica
ti
o
n
b
y
d
e
t
e
r
m
in
in
g
u
sa
b
le
s
p
e
e
c
h
,
J.
ACOU
S
T
.
S
OC.
AM
.
,
2
0
0
8
–
V
o
l
.
1
2
4
,
issu
e
5
.
[7
]
Bre
tt
Y
S
m
o
len
sk
i
a
n
d
Ra
v
i
P
.
Ra
m
a
c
h
a
n
d
ra
n
,
Us
a
b
le
S
p
e
e
c
h
p
r
o
c
e
ss
in
g
:
a
f
il
terle
s
s
a
p
p
ro
a
c
h
in
th
e
p
re
se
n
c
e
o
f
in
terf
e
re
n
c
e
;
IEE
E
Circ
u
it
s a
n
d
S
y
ste
ms
M
a
g
a
zin
e
,
(2
0
1
1
).
[8
]
K.R.
Krish
n
a
m
a
c
h
a
ri,
R.
E.
Ya
n
to
rn
o
,
D.S
.
Be
n
in
c
a
sa
a
n
d
S
.
J.
W
e
n
n
d
t,
“
S
p
e
c
tral
a
u
to
c
o
r
re
latio
n
ra
ti
o
a
s a u
sa
b
il
it
y
m
e
a
su
re
o
f
sp
e
e
c
h
se
g
m
e
n
ts
u
n
d
e
r
c
o
c
h
a
n
n
e
l
c
o
n
d
it
i
o
n
s
”
,
IEE
E
I
n
ter
n
a
ti
o
n
a
l
S
y
mp
o
siu
m
I
n
telli
g
e
n
t
S
ig
.
Pro
c
e
ss
.
a
n
d
Co
mm
S
y
s
.
,
(2
0
0
0
).
[9
]
Kiz
h
a
n
a
th
a
m
,
R
.
E.
Ya
n
to
rn
o
,
P
e
a
k
Diff
e
re
n
c
e
Au
to
c
o
rr
e
la
ti
o
n
o
f
W
a
v
e
let
T
ra
n
sfo
rm
Al
g
o
rit
h
m
B
a
se
d
Us
a
b
le
S
p
e
e
c
h
M
e
a
su
re
,
7
th
W
o
rl
d
M
u
lt
i
-
c
o
n
f
e
re
n
c
e
o
n
S
y
ste
m
ic,
C
y
b
e
rn
e
ti
c
s,
a
n
d
I
n
f
o
rm
a
ti
c
s,
2
0
0
3
.
[1
0
]
W
.
G
h
e
z
a
iel,
A
.
Be
n
S
li
m
a
n
e
,
E.
Be
n
Bra
iek
.
U
sa
b
le
sp
e
e
c
h
d
e
te
c
ti
o
n
f
o
r
sp
e
a
k
e
r
id
e
n
ti
f
ica
ti
o
n
sy
ste
m
u
n
d
e
r
c
o
-
c
h
a
n
n
e
l
c
o
n
d
i
ti
o
n
,
JT
EA
2
0
1
0
T
u
n
isia.
[1
1
]
W
a
jd
i
G
h
e
z
a
iel
,
Am
e
l
B
e
n
S
li
m
a
n
e
,
Ezz
e
d
in
e
Be
n
Bra
ie
k
.
E
v
a
lu
a
ti
o
n
o
f
a
m
u
lt
i
-
re
so
lu
ti
o
n
d
y
a
d
ic
wa
v
e
let
tran
sf
o
r
m
m
e
th
o
d
f
o
r
u
sa
b
le s
p
e
e
c
h
d
e
tec
ti
o
n
,
w
a
se
t
jo
u
rn
a
l
.
2
0
1
1
–
V
o
l.
7
9
–
P
.
8
2
9
-
8
3
3
.
[1
2
]
W
a
jd
i
G
h
e
z
a
iel
,
Am
e
l
B
e
n
S
li
m
a
n
e
,
Ezz
e
d
in
e
Be
n
Bra
iek
.
“
Us
a
b
le
S
p
e
e
c
h
A
s
sig
n
m
e
n
t
f
o
r
S
p
e
a
k
e
r
Id
e
n
ti
f
ica
ti
o
n
u
n
d
e
r
C
o
-
Ch
a
n
n
e
l
S
it
u
a
ti
o
n
”
.
In
t
e
rn
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
Co
mp
u
ter
Ap
p
li
c
a
ti
o
n
s
,
5
9
(
1
8
)
:
7
-
1
1
,
De
c
e
m
b
e
r
2
0
1
2
.
[1
3
]
J.
S
iri
g
o
s,
N.
F
a
k
o
tak
is,
G
.
Ko
k
k
in
a
k
is:
“
A
c
o
mp
a
riso
n
o
f
se
v
e
ra
l
sp
e
e
c
h
p
a
ra
me
ter
s
fo
r
sp
e
a
k
e
r
in
d
e
p
e
n
d
e
n
t
sp
e
e
c
h
re
c
o
g
n
it
i
o
n
a
n
d
s
p
e
a
k
e
r re
c
o
g
n
it
i
o
n
”
,
in
p
r
o
c
.
Eu
r
o
sp
e
e
c
h
„
9
5
,
M
a
d
ri
d
,
S
p
a
in
,
(1
9
9
5
)
1
8
-
21.
[1
4
]
S
h
a
o
Y.
a
n
d
W
a
n
g
D.
L
.
(2
0
0
3
):
Co
-
c
h
a
n
n
e
l
sp
e
a
k
e
r
id
e
n
ti
fi
c
a
ti
o
n
u
sin
g
u
sa
b
le
sp
e
e
c
h
e
x
tra
c
ti
o
n
b
a
se
d
o
n
mu
lt
i
-
p
it
c
h
tr
a
c
k
in
g
.
P
r
o
c
e
e
d
in
g
s o
f
ICA
S
S
P
-
0
3
,
v
o
l
.
I
I.
2
0
5
-
2
0
8
.
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
Wa
jd
i
G
h
e
z
a
iel
b
o
rn
in
T
u
n
is
(
T
u
n
isia),
h
e
re
c
e
iv
e
d
th
e
b
a
c
h
e
lo
r
d
e
g
re
e
f
ro
m
th
e
h
ig
h
sc
h
o
o
l
o
f
sc
ien
c
e
s
a
n
d
tec
h
n
iq
u
e
s
o
f
T
u
n
isia
si
n
c
e
2
0
0
1
,
th
e
M
a
ste
r
De
g
re
e
a
n
d
th
e
P
h
.
D
d
e
g
re
e
in
S
ig
n
a
l
p
r
o
c
e
ss
in
g
f
ro
m
Hig
h
S
c
h
o
o
l
o
f
S
c
ien
c
e
s
a
n
d
T
e
c
h
n
iq
u
e
s
o
f
T
u
n
is
re
sp
e
c
ti
v
e
l
y
in
2
0
0
4
a
n
d
2
0
1
4
.
He
b
e
lo
n
g
s
to
th
e
CERE
P
g
ro
u
p
in
t
h
e
h
ig
h
sc
h
o
o
l
o
f
sc
ien
c
e
s
a
n
d
tec
h
n
iq
u
e
s
o
f
T
u
n
isia.
Dr
G
h
e
z
a
iel
h
a
s
p
u
b
li
s
h
e
d
o
v
e
r
1
1
sc
h
o
larly
re
se
a
rc
h
p
a
p
e
rs
in
m
a
n
y
jo
u
rn
a
l
a
n
d
in
tern
a
ti
o
n
a
l
c
o
n
f
e
re
n
c
e
s
a
n
d
h
e
i
s
a
lrea
d
y
su
p
e
rv
isin
g
o
v
e
r
ten
m
a
ste
rs
a
n
d
e
n
g
in
e
e
r
a
p
p
li
c
a
ti
o
n
p
ro
jec
ts.
His res
e
a
rc
h
in
tere
sts
a
re
f
o
c
u
sin
g
o
n
sig
n
a
l,
v
o
ice
re
c
o
g
n
it
io
n
a
n
d
f
il
terin
g
.
A
m
e
l
B
e
n
S
li
m
a
n
e
is
c
u
rre
n
tl
y
As
sista
n
t
P
ro
f
e
ss
o
r
o
f
tele
c
o
m
m
u
n
ica
ti
o
n
s
a
t
th
e
Na
ti
o
n
a
l
S
c
h
o
o
l
o
f
Co
m
p
u
ter
S
c
ien
c
e
s
(Eco
le
Na
ti
o
n
a
le
d
e
s
S
c
ien
c
e
s
d
e
l'
In
f
o
rm
a
ti
q
u
e
,
ENS
I),
Un
iv
e
rsit
y
o
f
M
a
n
o
u
b
a
,
T
u
n
isia
sin
c
e
2
0
0
3
.
S
h
e
re
c
e
iv
e
d
h
e
r
En
g
in
e
e
rin
g
d
e
g
re
e
a
n
d
P
h
.
D
in
e
lec
tri
c
a
l
e
n
g
in
e
e
rin
g
b
o
th
f
ro
m
Na
ti
o
n
a
l
S
c
h
o
o
l
o
f
En
g
in
e
e
rin
g
o
f
T
u
n
is
(Eco
le
Na
ti
o
n
a
le
d
‟In
g
é
n
ieu
rs
d
e
T
u
n
is,
ENIT
)
re
sp
e
c
ti
v
e
l
y
in
1
9
8
5
a
n
d
2
0
0
3
.
H
e
r
re
se
a
rc
h
in
tere
sts
f
o
c
u
s
o
n
sig
n
a
l
p
ro
c
e
ss
in
g
.
P
a
rti
c
u
larly
,
S
h
e
w
o
rk
s o
n
sp
e
e
c
h
p
ro
c
e
ss
in
g
.
T
h
e
re
su
lt
s o
f
h
e
r
re
se
a
rc
h
w
o
rk
h
a
v
e
b
e
e
n
p
u
b
li
sh
e
d
i
n
m
a
n
y
in
tern
a
ti
o
n
a
l
c
o
n
f
e
re
n
c
e
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
6
,
No
.
6
,
Decem
b
er
201
6
: 2
7
6
6
–
2
7
7
2
2772
Ez
z
e
d
in
e
B
e
n
B
r
a
ie
k
o
b
tain
e
d
h
is
HD
R
o
n
2
0
0
8
in
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
f
ro
m
ENS
E
T
T
u
n
isia.
He
is,
p
re
se
n
tl
y
,
p
ro
fe
ss
o
r
in
t
h
e
d
e
p
a
rtm
e
n
t
o
f
e
l
e
c
tri
c
a
l
e
n
g
in
e
e
rin
g
a
t
th
e
tec
h
n
ica
l
uni
v
e
rsity
ES
S
T
T
a
n
d
m
a
n
a
g
e
r
o
f
th
e
re
se
a
rc
h
g
ro
u
p
o
n
sig
n
a
l
a
n
d
im
a
g
e
p
ro
c
e
ss
in
g
a
t
th
e
CERE
P
.
His
f
ield
s
o
f
in
tere
st
in
c
lu
d
e
a
u
t
o
m
a
ti
c
s,
e
le
c
tro
n
ics
,
c
o
n
tro
l
,
c
o
m
p
u
ter
v
isio
n
,
im
a
g
e
p
ro
c
e
ss
in
g
a
n
d
it
s ap
p
li
c
a
ti
o
n
in
h
a
n
d
w
rit
ten
d
a
ta rec
o
g
n
it
i
o
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.