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.
7
,
No
.
4
,
A
u
g
u
s
t
201
7
,
p
p
.
1
9
4
1
~
1
9
5
1
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v7
i
4
.
p
p
1
9
4
1
-
1951
1941
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
Sin
g
le Channel
S
peech Enha
nc
e
ment u
sing
Wiener
Filt
er
and
Co
m
pr
ess
iv
e Sen
sing
A
m
a
rt
S
ulo
ng
1
,
T
eddy
S
ury
a
G
un
a
w
a
n
2
,
O
t
h
m
a
n O
.
K
h
a
lifa
3
,
M
ira
K
a
rt
i
w
i
4
,
H
a
s
s
a
n Da
o
5
1,
2,
3
De
p
a
rtm
e
n
t
o
f
El
e
c
tri
c
a
l
a
n
d
Co
m
p
u
ter E
n
g
in
e
e
rin
g
,
I
n
tern
a
ti
o
n
a
l
Isla
m
ic Un
iv
e
rsit
y
M
a
l
a
y
sia
,
M
a
la
y
sia
4
De
p
a
rtme
n
t
o
f
In
f
o
rm
a
ti
o
n
S
y
ste
m
s,
In
tern
a
ti
o
n
a
l
Isla
m
ic Un
iv
e
rs
it
y
M
a
la
y
sia
,
M
a
la
y
sia
5
In
stit
u
te
o
f
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
lo
g
y
,
Un
iv
e
rsit
y
Ku
a
la L
u
m
p
u
r,
M
a
la
y
sia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
an
4
,
2
0
1
7
R
ev
i
s
ed
Ma
y
3
1
,
2
0
1
7
A
cc
ep
ted
J
u
n
1
4
,
2
0
1
7
T
h
e
sp
e
e
c
h
e
n
h
a
n
c
e
m
e
n
t
a
lg
o
rit
h
m
s
a
re
u
ti
li
z
e
d
to
o
v
e
rc
o
m
e
m
u
lt
ip
le
li
m
it
a
ti
o
n
f
a
c
to
rs
in
re
c
e
n
t
a
p
p
li
c
a
ti
o
n
s
su
c
h
a
s
m
o
b
il
e
p
h
o
n
e
a
n
d
c
o
m
m
u
n
ica
ti
o
n
c
h
a
n
n
e
l.
T
h
e
c
h
a
ll
e
n
g
e
s
f
o
c
u
s
o
n
c
o
rru
p
ted
sp
e
e
c
h
so
lu
t
io
n
b
e
tw
e
e
n
n
o
ise
re
d
u
c
ti
o
n
a
n
d
sig
n
a
l
d
isto
r
ti
o
n
.
W
e
u
se
d
a
m
o
d
if
i
e
d
W
ien
e
r
f
il
ter
a
n
d
c
o
m
p
re
ss
iv
e
se
n
sin
g
(CS
)
to
i
n
v
e
stig
a
te
a
n
d
e
v
a
lu
a
te
th
e
im
p
ro
v
e
m
e
n
t
o
f
sp
e
e
c
h
q
u
a
li
ty
.
T
h
is
n
e
w
m
e
th
o
d
a
d
a
p
ted
n
o
ise
e
sti
m
a
ti
o
n
a
n
d
W
ien
e
r
f
il
ter
g
a
in
f
u
n
c
ti
o
n
in
w
h
ich
to
i
n
c
re
a
se
we
ig
h
t
a
m
p
li
tu
d
e
sp
e
c
tru
m
a
n
d
i
m
p
ro
v
e
m
it
ig
a
ti
o
n
o
f
in
tere
ste
d
sig
n
a
ls.
T
h
e
CS
is
th
e
n
a
p
p
li
e
d
u
si
n
g
th
e
g
ra
d
ien
t
p
r
o
jec
ti
o
n
f
o
r
sp
a
rse
re
c
o
n
str
u
c
ti
o
n
(G
P
S
R)
tec
h
n
iq
u
e
a
s
a
stu
d
y
s
y
ste
m
to
e
m
p
iri
c
a
ll
y
in
v
e
stig
a
te
th
e
in
tera
c
t
iv
e
e
ff
e
c
ts
o
f
th
e
c
o
rru
p
ted
n
o
ise
a
n
d
o
b
tai
n
b
e
tt
e
r
p
e
rc
e
p
tu
a
l
im
p
ro
v
e
m
e
n
t
a
sp
e
c
ts
to
li
ste
n
e
r
f
a
ti
g
u
e
w
it
h
n
o
ise
les
s
re
d
u
c
ti
o
n
c
o
n
d
it
io
n
s.
T
h
e
p
ro
p
o
se
d
a
lg
o
rit
h
m
sh
o
w
s
a
n
e
n
h
a
n
c
e
m
e
n
t
in
tes
ti
n
g
p
e
rf
o
rm
a
n
c
e
e
v
a
lu
a
ti
o
n
o
f
o
b
jec
ti
v
e
a
ss
e
ss
m
e
n
t
tes
ts
o
u
tp
e
rf
o
rm
c
o
m
p
a
re
d
to
o
th
e
r
c
o
n
v
e
n
ti
o
n
a
l
a
lg
o
rit
h
m
s
a
t
v
a
rio
u
s
n
o
ise
ty
p
e
c
o
n
d
it
i
o
n
s
o
f
0
,
5
,
1
0
,
1
5
d
B
S
NRs
.
T
h
e
re
f
o
re
,
th
e
p
ro
p
o
se
d
a
lg
o
rit
h
m
sig
n
if
ic
a
n
tl
y
a
c
h
iev
e
d
th
e
sp
e
e
c
h
q
u
a
li
ty
i
m
p
ro
v
e
m
e
n
t
a
n
d
e
f
f
icie
n
tl
y
o
b
tain
e
d
h
ig
h
e
r
p
e
rf
o
rm
a
n
c
e
re
su
lt
in
g
i
n
b
e
tt
e
r
n
o
ise
re
d
u
c
ti
o
n
c
o
m
p
a
re
to
o
th
e
r
c
o
n
v
e
n
ti
o
n
a
l
a
lg
o
rit
h
m
s.
K
ey
w
o
r
d
:
C
o
m
p
r
ess
iv
e
s
e
n
s
in
g
P
E
SQ
P
E
SQ i
m
p
r
o
v
e
m
e
n
t
SNR
Sp
ee
ch
en
h
an
ce
m
e
n
t
W
ien
er
f
ilter
Co
p
y
rig
h
t
©
2
0
1
7
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
:
T
ed
d
y
S
u
r
y
a
G
u
n
a
w
an
,
Dep
ar
t
m
en
t o
f
E
lectr
ical
an
d
C
o
m
p
u
ter
E
n
g
in
ee
r
i
n
g
,
I
n
ter
n
atio
n
al
I
s
la
m
ic
U
n
iv
er
s
it
y
Ma
la
y
s
ia,
J
alan
Go
m
b
ak
,
5
3
1
0
0
Ku
ala
L
u
m
p
u
r
,
Ma
la
y
s
ia.
E
m
ail: t
s
g
u
n
a
w
an
@
i
iu
m
.
ed
u
.
m
y
1.
I
NT
RO
D
UCT
I
O
N
I
n
th
e
m
o
d
er
n
er
a,
th
e
ad
v
a
n
c
e
m
en
t
in
tec
h
n
o
lo
g
y
h
as
b
r
o
u
g
h
t
ab
o
u
t
g
r
ea
t
b
en
e
f
its
to
h
u
m
an
b
ei
n
g
s
an
d
th
eir
d
ail
y
li
v
e
s
.
T
o
d
ay
i
n
n
o
v
atio
n
tec
h
n
o
lo
g
ies,
s
ig
n
a
l
p
r
o
ce
s
s
in
g
i
s
o
n
e
o
f
t
h
e
m
o
s
t
p
o
w
er
f
u
l
s
o
u
r
ce
s
f
o
r
m
o
d
er
n
d
esig
n
ed
en
g
in
ee
r
in
g
t
h
at
ca
p
ab
le
o
f
r
ea
lizin
g
v
ar
io
u
s
ap
p
licatio
n
s
i
n
th
eir
r
ea
l
i
m
p
le
m
e
n
tat
io
n
s
f
r
o
m
t
h
eo
r
etica
l
a
s
p
ec
t
p
o
in
t
o
f
v
ie
w
to
co
u
n
ter
p
ar
t
i
n
d
if
f
er
en
t
ap
p
licatio
n
ar
ea
s
.
T
h
er
e
ar
e
al
w
a
y
s
a
tr
ad
e
-
o
f
f
b
et
w
ee
n
n
o
i
s
e
r
ed
u
ctio
n
a
n
d
s
ig
n
al
d
is
to
r
tio
n
.
Mo
s
t
o
f
r
esear
ch
f
o
u
n
d
t
h
at
m
o
r
e
n
o
is
e
r
ed
u
ctio
n
is
al
w
a
y
s
ac
co
m
p
a
n
ied
b
y
m
o
r
e
s
i
g
n
al
d
is
to
r
tio
n
[
1
]
,
[
2
]
.
T
h
e
m
ai
n
c
h
alle
n
g
e
o
f
th
e
s
p
ee
ch
en
h
a
n
c
e
m
en
t
p
r
o
ce
s
s
i
s
to
d
esig
n
ef
f
ec
ti
v
e
al
g
o
r
ith
m
s
to
s
u
p
p
r
ess
t
h
e
n
o
i
s
e
w
it
h
o
u
t
i
n
tr
o
d
u
cin
g
a
n
y
p
o
s
s
ib
ilit
y
o
f
p
er
ce
p
tu
al
d
is
to
r
tio
n
in
to
t
h
e
s
p
ee
ch
s
i
g
n
a
l
[
1
]
,
[
3
]
.
R
esear
ch
an
d
i
n
v
esti
g
atio
n
s
o
n
s
p
ee
ch
e
n
h
an
ce
m
e
n
t
p
r
o
b
lem
h
a
v
e
b
ee
n
g
r
o
w
i
n
g
at
a
r
ap
id
r
at
e
t
h
at
c
o
v
er
a
b
r
o
ad
s
p
ec
tr
u
m
o
f
co
n
s
tr
ain
s
,
ap
p
licatio
n
,
a
n
d
i
s
s
u
e
s
.
T
h
e
ch
al
len
g
i
n
g
w
o
r
k
f
o
r
en
h
a
n
ci
n
g
n
o
is
y
s
p
ee
ch
is
o
n
s
i
n
g
le
m
icr
o
p
h
o
n
e
a
n
d
th
e
s
p
ee
ch
p
r
o
b
le
m
th
at
was
d
eg
r
ad
ed
b
y
th
e
n
o
is
e
an
d
r
e
m
ain
s
w
id
el
y
o
p
en
f
o
r
in
v
e
s
ti
g
atio
n
[
3
]
,
[
4
]
.
Su
c
h
p
r
o
b
lem
i
s
w
ell
k
n
o
w
n
as
s
in
g
le
-
c
h
an
n
el
s
p
ee
ch
en
h
an
ce
m
e
n
t
an
d
co
n
s
id
er
ed
as
th
e
m
o
s
t
d
if
f
ic
u
lt
ta
s
k
[
1
]
,
[
5
]
.
T
h
is
is
b
ec
au
s
e
o
f
f
ac
t
th
at
th
e
n
o
is
e
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.
7
,
No
.
4
,
A
u
g
u
s
t
2017
:
1
9
4
1
–
1
9
5
1
1942
an
d
s
p
ee
ch
ar
e
p
er
ce
iv
ed
as
w
it
h
i
n
t
h
e
s
a
m
e
ch
a
n
n
el
b
y
a
s
s
u
m
in
g
n
o
ac
ce
s
s
to
r
ef
er
en
ce
n
o
is
e
w
h
er
e
t
h
e
i
m
p
r
o
v
e
m
en
t o
f
th
e
s
p
ee
ch
s
i
g
n
al
-
to
-
n
o
is
e
r
atio
(
SNR
)
i
s
tar
g
et
o
f
m
o
s
t te
c
h
n
iq
u
es.
Mo
s
t
o
f
t
h
e
s
p
ee
ch
e
n
h
an
ce
m
en
t
tech
n
iq
u
es
h
av
e
co
n
ce
n
tr
ated
p
r
in
cip
all
y
o
n
s
tatis
ticall
y
u
n
co
r
r
elate
d
an
d
i
n
d
ep
en
d
en
t
ad
d
itiv
e
n
o
is
e
[
3
]
,
[
5
]
.
Ho
w
ev
er
,
th
e
d
esi
g
n
o
f
e
f
f
ec
tiv
e
al
g
o
r
ith
m
s
th
at
ca
n
co
m
b
at
ad
d
itiv
e
n
o
is
e
w
h
ile
p
r
o
d
u
cin
g
h
ig
h
q
u
ali
t
y
a
n
d
i
m
p
r
o
v
ed
s
p
ee
ch
s
ig
n
al
is
li
m
ite
d
.
T
h
u
s
,
th
e
s
tu
d
ie
s
o
f
ad
d
itiv
e
n
o
is
e
i
n
v
ar
io
u
s
t
y
p
es
o
f
ap
p
licatio
n
s
an
d
th
eir
r
elate
d
b
eh
av
io
r
ar
e
cr
u
cial
en
d
ea
v
o
r
s
.
Mo
s
t
o
f
th
e
liter
atu
r
es
f
o
cu
s
o
n
t
h
e
d
if
f
er
en
ce
o
f
t
h
e
n
o
i
s
e
s
o
u
r
ce
s
in
t
er
m
s
o
f
te
m
p
o
r
al
an
d
s
p
ec
tr
al
ch
ar
ac
ter
is
tic
s
,
an
d
th
e
r
a
n
g
e
o
f
t
h
e
n
o
is
e
lev
el
s
th
at
m
a
y
b
e
e
n
co
u
n
ter
ed
in
r
ea
l
lif
e
[
1
]
.
Ma
n
y
ex
is
ti
n
g
r
e
s
ea
r
ch
es
o
n
s
p
ee
ch
en
h
a
n
ce
m
en
t
h
av
e
b
ased
r
elati
v
el
y
o
n
s
a
m
p
les
o
f
s
p
ee
c
h
q
u
a
lit
y
m
ea
s
u
r
e
m
e
n
ts
w
h
ic
h
h
as
m
ad
e
it
i
m
p
o
s
s
ib
le
to
ca
r
r
y
o
u
t
s
ati
s
f
ac
to
r
y
s
tu
d
ies.
T
h
is
asp
ec
t
o
f
s
tu
d
y
m
a
y
s
u
g
g
est
a
b
etter
u
n
d
er
s
ta
n
d
in
g
o
f
t
h
e
r
elate
d
ch
ar
ac
ter
is
tic
s
w
i
th
a
g
r
ea
t
n
u
m
b
er
o
f
th
e
n
o
i
s
y
s
p
ee
c
h
d
ate
av
ailab
le
f
o
r
th
e
s
p
ee
ch
at
v
ar
io
u
s
d
B
SNR
en
v
ir
o
n
m
e
n
t
s
[
1
]
.
C
o
n
ce
r
n
s
h
a
v
e
b
ee
n
e
x
p
r
ess
e
d
ab
o
u
t
s
p
ee
ch
en
h
an
ce
m
e
n
t
ap
p
r
o
ac
h
es.
Ho
w
ev
er
,
t
h
er
e
h
as
b
ee
n
a
f
e
w
r
esear
c
h
es
s
o
f
ar
t
h
at
s
ee
k
p
o
s
s
ib
le
s
o
l
u
tio
n
to
th
e
s
p
ee
ch
en
h
a
n
ce
m
e
n
t
b
ased
o
n
co
m
p
r
e
s
s
i
v
e
s
e
n
s
in
g
(
C
S)
tech
n
iq
u
e.
C
o
n
s
eq
u
e
n
tl
y
,
th
e
q
u
esti
o
n
r
e
m
ai
n
w
h
e
th
er
it
ca
n
ac
h
iev
e
s
u
itab
le
h
ig
h
i
m
p
r
o
v
e
m
e
n
t
i
n
b
o
th
its
p
er
f
o
r
m
a
n
ce
an
d
q
u
alit
y
.
T
h
u
s
,
it
m
a
y
b
e
u
s
e
f
u
l
to
in
v
esti
g
ate
a
n
d
an
al
y
ze
t
h
i
s
n
e
w
ap
p
r
o
ac
h
o
f
d
ata
ac
q
u
is
itio
n
w
h
ic
h
is
k
n
o
w
n
a
s
co
m
p
r
ess
iv
e
s
e
n
s
i
n
g
(
C
S)
tec
h
n
iq
u
e
[
6
]
.
I
ts
th
eo
r
y
ass
er
t
t
h
at
o
n
e
ca
n
r
ec
o
v
er
ce
r
tain
s
i
g
n
al
s
f
r
o
m
f
ar
f
e
w
er
s
a
m
p
les
o
r
m
ea
s
u
r
e
m
e
n
t
s
th
a
n
co
n
v
e
n
tio
n
al
m
et
h
o
d
th
at
is
b
ased
o
n
th
e
w
ell
-
k
n
o
w
n
Sh
a
n
n
o
n
/N
y
q
u
is
t
s
a
m
p
lin
g
t
h
eo
r
e
m
[
7
,
8
]
.
I
n
t
u
r
n
,
n
e
w
t
y
p
e
o
f
s
a
m
p
li
n
g
t
h
eo
r
y
ca
n
p
r
ed
ict
f
r
o
m
th
e
s
p
ar
ce
s
ig
n
als
a
n
d
b
e
co
n
s
tr
u
cted
f
r
o
m
w
h
at
p
r
ev
io
u
s
l
y
b
eliev
ed
to
b
e
in
co
m
p
lete
i
n
f
o
r
m
atio
n
[
6
]
.
T
h
is
m
et
h
o
d
also
p
r
o
v
id
es
ef
f
icie
n
t
alg
o
r
it
h
m
w
h
ic
h
ca
n
b
e
u
s
ed
f
o
r
p
er
f
ec
t
r
ec
o
v
er
y
o
f
t
h
e
s
p
ar
s
e
s
ig
n
al
[
9
]
.
Ma
j
o
r
ity
o
f
r
esear
ch
e
s
i
n
t
h
e
C
S tec
h
n
iq
u
e
s
h
av
e
b
ee
n
in
tr
o
d
u
c
ed
in
i
m
a
g
e
p
r
o
ce
s
s
in
g
to
p
r
o
v
id
e
co
m
p
r
es
s
ed
v
er
s
io
n
o
f
th
e
o
r
ig
i
n
al
i
m
ag
e
w
it
h
n
o
i
s
eles
s
d
is
to
r
tio
n
[
6
,
9
]
.
T
h
is
tech
n
iq
u
e
r
elie
s
m
ain
l
y
o
n
e
m
p
ir
ica
l
o
b
s
er
v
atio
n
th
a
t
m
a
n
y
s
i
g
n
a
ls
ca
n
b
e
w
ell
-
ap
p
r
o
x
i
m
ated
b
y
s
p
ar
s
e
ex
p
r
ess
io
n
i
n
ter
m
s
o
f
s
u
itab
le
b
asis
[
6
].
2.
L
I
T
E
R
AT
U
RE
SU
RVE
Y
O
F
SPEE
CH
E
NH
ANC
E
M
E
NT
Ma
n
y
liter
at
u
r
es
h
a
v
e
b
ee
n
r
ep
o
r
t
[
1
]
,
[
3
]
,
[
11
]
,
[
2
8
]
an
d
m
en
tio
n
ed
a
w
id
el
y
u
s
ed
s
i
n
g
le
c
h
an
n
el
s
p
ee
ch
en
h
an
ce
m
e
n
t
b
ased
o
n
th
e
s
h
o
r
t
-
ti
m
e
s
p
ec
tr
al
m
ag
n
it
u
d
e
(
ST
SM)
.
I
n
r
ea
l
p
r
o
ce
s
s
in
g
s
p
ee
c
h
en
h
a
n
ce
m
en
t
tech
n
iq
u
e
s
,
th
e
alg
o
r
ith
m
e
m
p
lo
y
ed
a
s
i
m
p
l
e
p
r
in
cip
le
in
w
h
ich
t
h
e
s
p
e
ctr
u
m
o
f
t
h
e
clea
n
s
p
ee
ch
esti
m
atio
n
s
i
g
n
al
ca
n
b
e
o
b
tain
ed
b
y
s
u
b
tr
ac
tin
g
a
n
o
is
e
esti
m
at
io
n
s
p
ec
tr
u
m
f
r
o
m
th
e
n
o
is
y
s
p
ee
c
h
s
p
ec
tr
u
m
co
n
d
itio
n
s
.
I
n
g
e
n
er
al,
s
p
ee
ch
en
h
a
n
ce
m
e
n
t
[
1
]
,
[
1
2
]
w
a
s
co
n
ta
m
i
n
ated
an
d
d
eg
r
ad
ed
w
ith
ad
d
iti
v
e
n
o
is
e.
I
t
is
t
y
p
icall
y
attac
k
ed
b
y
th
e
b
ac
k
g
r
o
u
n
d
n
o
is
e
o
f
u
n
co
r
r
elate
d
s
p
ee
ch
.
T
h
is
s
ig
n
al
w
a
s
k
n
o
w
n
as
n
o
is
y
s
p
ee
ch
an
d
its
s
p
ec
tr
u
alr
u
m
ca
n
b
e
ex
p
r
ess
ed
as f
o
llo
w
;
)
(
)
(
)
(
n
d
n
s
n
y
an
d
)
,
(
)
,
(
)
,
(
k
D
k
S
k
Y
(
1
)
w
h
er
e
)
(
n
y
,
)
(
n
s
,
an
d
)
(
n
d
ar
e
n
o
is
y
s
p
ee
ch
,
clea
n
s
p
ee
ch
,
an
d
ad
d
itiv
e
n
o
is
e
r
esp
ec
tiv
el
y
w
i
th
n
s
a
m
p
le
n
u
m
b
er
o
f
t
h
e
d
is
cr
ete
ti
m
e
s
i
g
n
al.
I
t
i
s
o
f
te
n
co
m
p
u
ted
o
n
a
f
r
a
m
e
-
by
-
f
r
a
m
e
b
as
is
.
T
h
e
n
o
is
y
s
p
ee
c
h
is
t
h
e
n
ca
lcu
lated
i
n
th
e
d
is
cr
ete
t
i
m
e
d
o
m
ain
o
f
th
e
s
h
o
r
t
-
t
i
m
e
F
o
u
r
ier
tr
an
s
f
o
r
m
(
ST
FT
)
in
wh
ich
it
is
g
e
n
er
all
y
n
o
n
-
s
tat
io
n
ar
y
in
n
at
u
r
e
.
I
ts
n
o
is
y
s
p
ee
ch
s
p
ec
tr
u
m
)
,
(
k
Y
,
clea
n
s
p
ee
c
h
s
p
ec
tr
u
m
)
,
(
k
S
,
an
d
n
o
i
s
e
s
p
ec
tr
u
m
)
,
(
k
D
ar
e
ca
lcu
lated
d
ep
en
d
o
n
an
d
k
(
an
d
k
ar
e
d
en
o
ted
as
f
r
eq
u
e
n
c
y
r
esp
o
n
s
e
an
d
t
h
e
f
r
a
m
e
n
u
m
b
er
r
esp
ec
ti
v
el
y
)
.
Fo
r
s
i
m
p
licit
y
,
t
h
e
k
ter
m
th
r
o
u
g
h
o
u
t
th
e
a
s
s
u
m
p
t
io
n
o
f
a
f
r
a
m
e
s
e
g
m
en
t
ar
e
d
r
o
p
p
ed
.
Hen
ce
th
e
n
o
is
y
s
p
ee
ch
p
o
w
er
s
p
ec
tr
u
m
ca
n
b
e
ex
p
r
ess
ed
as f
o
llo
w
s
2
2
2
)
(
)
(
)
(
D
S
Y
(
2
)
T
h
e
en
h
an
ce
d
s
p
ee
c
h
esti
m
ati
o
n
in
s
h
o
r
t
-
t
i
m
e
m
a
g
n
itu
d
e
)
(
ˆ
S
ca
n
b
e
o
b
tain
b
y
s
u
b
tr
ac
ti
n
g
a
n
o
is
e
esti
m
ate
d
u
r
in
g
s
p
ee
c
h
p
au
s
e,
w
h
ic
h
f
o
r
m
u
lated
as
f
o
llo
w
:
0
)
(
ˆ
)
(
)
(
ˆ
)
(
)
(
ˆ
2
2
2
2
2
D
Y
D
Y
S
if
(3
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
S
in
g
le
C
h
a
n
n
el
S
p
ee
c
h
E
n
h
a
n
ce
men
t u
s
in
g
W
ien
er F
ilte
r
a
n
d
C
o
mp
r
ess
ive
S
en
s
in
g
(
A
ma
r
t S
u
lo
n
g
)
1943
T
h
e
n
o
is
e
esti
m
atio
n
s
p
ec
tr
u
m
2
)
(
ˆ
D
is
ca
lcu
lated
f
r
o
m
t
h
e
av
er
ag
i
n
g
f
r
a
m
e
s
o
f
t
h
e
r
ec
en
t sp
ee
c
h
p
au
s
es:
1
0
2
2
)
(
ˆ
1
)
(
ˆ
M
j
SP
Y
M
D
(4
)
w
h
er
e
M
is
t
h
e
n
u
m
b
er
o
f
s
p
ee
ch
p
au
s
es
in
co
n
s
ec
u
ti
v
e
f
r
a
m
es.
E
q
u
atio
n
(
4
)
is
n
o
t
ta
k
e
n
i
n
to
ac
tio
n
w
h
en
th
e
b
ac
k
g
r
o
u
n
d
n
o
i
s
e
is
s
tati
o
n
ar
y
an
d
co
v
er
ag
e
to
o
p
ti
m
al
esti
m
a
te
o
f
n
o
is
e
p
o
w
er
s
p
ec
tr
u
m
.
I
n
ad
d
itio
n
,
E
q
u
atio
n
(
3
)
ca
n
also
b
e
co
n
s
i
d
er
as f
ilter
w
h
e
n
its
p
r
o
d
u
ct
o
f
th
e
n
o
is
y
s
p
ee
ch
s
p
ec
tr
u
m
is
r
ep
r
esen
ted
as
2
2
2
2
)
(
)
(
)
(
ˆ
1
)
(
ˆ
Y
Y
D
S
2
)
(
)
(
Y
an
d
2
1
2
2
)
(
)
(
ˆ
1
,
0
m
a
x
)
(
Y
D
(5
)
w
h
er
e
)
(
is
th
e
g
a
in
f
u
n
ctio
n
o
f
s
p
ec
tr
al
s
u
b
tr
ac
tio
n
a
n
d
al
s
o
k
n
o
w
n
as
f
i
lter
.
T
h
is
g
ai
n
f
u
n
c
t
io
n
)
(
is
d
ef
in
ed
as
th
e
m
ag
n
it
u
d
e
r
esp
o
n
s
e
o
f
1
)
(
0
,
th
er
ef
o
r
e
it
is
ze
r
o
p
h
ase
f
ilter
as
s
h
o
w
n
)
(
in
E
q
u
atio
n
(6
)
.
T
o
s
y
n
th
e
s
is
e
r
esu
lt
s
,
th
e
e
n
h
a
n
ce
d
s
p
ee
ch
s
i
g
n
al
n
ee
d
s
r
ec
o
n
s
tr
u
ctio
n
.
T
h
is
p
h
ase
is
d
o
n
e
b
y
u
s
i
n
g
th
e
n
o
i
s
y
p
h
a
s
e
as
th
e
cl
ea
n
s
p
ee
ch
esti
m
at
io
n
s
i
g
n
al,
d
u
e
to
in
s
en
s
iti
v
it
y
o
f
t
h
e
h
u
m
an
au
d
ito
r
y
s
y
s
te
m
[5
]
,
[
1
3
]
.
Su
b
s
eq
u
en
t
l
y
,
t
h
e
e
n
h
a
n
ce
d
s
p
ee
c
h
i
n
a
f
r
a
m
e
is
esti
m
ated
an
d
t
h
e
clea
n
s
p
ee
ch
est
i
m
a
tio
n
is
t
h
e
n
s
y
n
t
h
esi
s
as
))
(
(
)
(
ˆ
)
(
ˆ
Y
j
e
S
I
F
F
T
n
s
.
I
t s
y
n
t
h
esi
s
w
il
l
r
ec
o
v
er
e
s
p
ee
ch
esti
m
at
io
n
w
a
v
ef
o
r
m
b
y
i
n
v
er
s
e
Fas
t
Fo
u
r
ier
tr
an
s
f
o
r
m
i
n
g
)
(
I
F
F
T
u
s
in
g
an
o
v
er
lap
an
d
ad
d
m
eth
o
d
.
Mo
r
eo
v
er
,
th
e
s
u
b
tr
ac
tiv
e
-
t
y
p
e
alg
o
r
ith
m
s
ca
n
also
b
e
e
s
ti
m
at
ed
u
s
i
n
g
f
ilter
ap
p
r
o
ac
h
d
ep
en
d
en
t
o
n
th
e
n
o
is
y
s
p
ee
ch
’
s
ch
a
r
ac
ter
is
tics
a
n
d
o
n
th
e
n
o
is
e
e
s
ti
m
atio
n
s
p
ec
tr
u
m
t
h
at
ca
n
b
e
e
x
p
r
ess
ed
a
s
)
(
)
(
)
(
ˆ
Y
S
.
T
h
i
s
g
ain
f
u
n
ctio
n
)
(
co
m
b
i
n
e
th
e
n
o
is
e
r
ed
u
ctio
n
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
[
1
4
]
w
it
h
[
1
5
]
.
I
n
ex
ten
s
iv
e
s
tu
d
ied
[
1
6
]
an
d
[
1
7
]
r
ep
o
r
ted
th
at
th
e
g
a
in
i
m
p
r
o
v
e
m
en
t
r
elati
v
el
y
u
s
ed
th
e
p
ar
a
m
eter
s
(
i.e
.
,
,
,
an
d
r
esp
ec
tiv
el
y
)
.
T
h
e
f
o
llo
w
in
g
g
ain
f
u
n
ct
io
n
is
a
s
f
o
llo
w
:
o
t
h
e
r
w
i
s
e
Y
D
Y
D
if
Y
D
2
1
1
2
1
)
(
)
(
ˆ
1
)
(
)
(
ˆ
,
)
(
)
(
ˆ
1
)
(
(6
)
Th
e
g
ain
f
u
n
c
tio
n
f
r
o
m
E
q
u
atio
n
(
6
)
is
th
e
d
esig
n
ed
p
ar
a
m
eter
to
d
ea
l
w
ith
t
h
e
tr
ad
eo
f
f
i
n
n
o
is
e
r
ed
u
ctio
n
,
r
esid
u
al
n
o
is
e
an
d
s
p
ee
ch
d
is
to
r
tio
n
s
i
g
n
a
l.
T
h
ese
v
ar
iatio
n
p
ar
a
m
eter
ca
n
b
e
d
escr
ib
ed
as
th
e
f
r
e
e
p
ar
am
eter
an
d
ca
n
b
e
d
escr
ib
ed
as
f
o
llo
w
s
:
a)
.
Ov
er
-
s
u
b
tr
a
ctio
n
f
ac
to
r
)
0
(
:
to
av
o
id
t
h
e
atte
n
u
at
io
n
o
f
th
e
s
p
ec
tr
u
m
m
o
r
e
th
a
n
n
ec
es
s
ar
y
w
h
ich
lead
s
to
t
h
e
r
ed
u
ce
d
r
esid
u
al
n
o
is
e
p
ea
k
s
ev
e
n
th
o
u
g
h
t
h
e
d
is
to
r
tio
n
to
t
h
e
s
p
ee
c
h
s
i
g
n
al
i
n
cr
ea
s
ed
(
r
e
d
u
cin
g
a
u
d
itab
le
d
is
to
r
tio
n
)
.
;
b
)
.
Sp
ec
tr
al
f
lo
o
r
in
g
)
1
0
(
:
to
r
ed
u
ce
th
e
b
ac
k
g
r
o
u
n
d
n
o
is
e
w
h
er
ea
s
t
h
e
b
ac
k
g
r
o
u
n
d
n
o
is
e
is
ad
d
ed
b
u
t
o
n
l
y
r
e
m
ai
n
in
g
th
e
m
i
n
i
m
u
m
v
al
u
e
o
f
th
e
b
ac
k
g
r
o
u
n
d
n
o
is
e
to
b
e
ta
k
en
.
; c
)
.
E
x
p
o
n
en
t
1
an
d
2
:
to
d
eter
m
i
n
e
th
e
s
h
ar
p
n
e
s
s
o
f
th
e
tr
an
s
itio
n
f
r
o
m
c
h
an
g
i
n
g
t
h
e
g
ain
f
u
n
ctio
n
,
b
y
as
s
i
g
n
i
n
g
1
)
(
(
m
o
d
i
f
ied
s
p
ec
tr
al
co
m
p
o
n
en
t)
.
T
h
e
m
o
d
i
f
icat
io
n
o
f
th
e
e
x
p
o
n
en
t
1
an
d
2
p
ar
am
eter
s
o
f
t
h
e
alg
o
r
i
th
m
a
n
d
its
r
es
u
lts
ar
e
d
escr
ib
ed
as
f
o
llo
w
:
in
ca
s
e
o
f
m
ag
n
it
u
d
e
s
u
b
tr
ac
tio
n
(
1
1
an
d
1
2
)
,
in
ca
s
e
o
f
p
o
w
e
r
s
p
ec
tr
al
s
u
b
tr
ac
tio
n
(
2
1
an
d
5
.
0
2
)
,
an
d
in
c
ase
o
f
W
ien
er
f
ilter
(
2
1
an
d
1
2
)
r
esp
ec
tiv
el
y
.
I
n
[
5
]
,
it
is
m
e
n
tio
n
ed
th
e
ad
v
an
ta
g
e
s
o
f
t
h
e
s
p
ec
tr
al
s
u
b
tr
ac
tio
n
al
g
o
r
ith
m
s
a
s
f
o
llo
w
;
1
)
s
i
m
p
le
a
n
d
o
n
l
y
r
eq
u
ir
in
g
n
o
is
e
es
ti
m
a
tio
n
s
p
ec
tr
u
m
,
a
n
d
2
)
v
ar
iatio
n
o
f
s
u
b
t
r
ac
tio
n
p
ar
a
m
eter
s
w
it
h
h
i
g
h
l
y
f
le
x
ib
ili
t
y
.
No
r
m
a
ll
y
,
it
e
m
p
lo
y
s
v
o
ice
a
ctiv
it
y
d
etec
tio
n
(
V
A
D)
in
t
h
e
f
o
r
m
o
f
s
tati
s
tical
in
f
o
r
m
a
tio
n
o
f
s
i
len
ce
r
eg
io
n
.
VAD
p
er
f
o
r
m
a
n
ce
d
eg
r
ad
ed
s
ig
n
if
ican
t
l
y
at
lo
w
s
i
g
n
al
to
n
o
is
e
r
atio
n
(
SNR
)
.
Ho
w
ev
er
,
d
if
f
ic
u
lt
y
e
m
er
g
e
d
w
h
e
n
b
ac
k
g
r
o
u
n
d
n
o
is
e
i
s
n
o
n
s
tat
io
n
ar
y
.
T
h
eir
s
h
o
r
tco
m
i
n
g
p
er
ce
p
tu
all
y
co
n
tai
n
s
t
h
e
r
e
m
n
a
n
t
o
f
u
n
n
at
u
r
al
n
o
ticea
b
le
to
s
p
ec
tr
al
ar
ti
f
ac
ts
k
n
o
w
n
as
m
u
s
ical
n
o
is
e
i
n
r
an
d
o
m
f
r
eq
u
e
n
cies.
I
t
co
r
r
ec
tl
y
d
ep
en
d
s
o
n
p
r
ec
is
e
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.
7
,
No
.
4
,
A
u
g
u
s
t
2017
:
1
9
4
1
–
1
9
5
1
1944
o
f
n
o
is
e
e
s
ti
m
atio
n
w
h
ic
h
i
s
li
m
ited
b
y
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
s
p
ee
ch
/p
a
u
s
e
d
etec
to
r
s
.
T
h
e
alg
o
r
ith
m
’
s
i
m
p
r
o
v
e
m
en
t
o
f
u
s
in
g
s
p
ec
tr
a
l
o
v
er
-
s
u
b
tr
ac
tio
n
is
to
m
i
n
i
m
ize
th
e
in
e
v
it
ab
le
n
o
is
e
an
d
d
is
to
r
tio
n
[
5
]
,
[
1
6
]
.
B
esid
e
th
at,
th
e
alg
o
r
ith
m
as
s
ig
n
ed
1
0
an
d
1
in
w
h
ich
to
co
n
tr
o
l
th
e
a
m
o
u
n
t
o
f
p
o
w
er
s
p
ec
tr
u
m
i
n
n
o
is
e
s
u
b
tr
ac
tio
n
f
r
o
m
th
e
p
o
w
er
s
p
ec
tr
u
m
o
f
th
e
n
o
is
y
s
p
ee
ch
in
ea
ch
s
p
ee
ch
f
r
a
m
e
[
5
]
.
I
ts
s
p
ec
tr
al
f
lo
o
r
p
ar
am
e
ter
s
ar
e
u
s
ed
to
p
r
ev
en
t
th
e
ca
u
s
e
o
f
s
p
ec
tr
u
m
f
lo
o
r
f
r
o
m
g
o
i
n
g
to
b
elo
w
t
h
e
p
r
eset
m
i
n
i
m
u
m
lev
el
r
ath
er
t
h
at
s
et
tin
g
to
ze
r
o
.
T
h
is
al
g
o
r
ith
m
d
ep
en
d
s
o
n
a
p
o
s
ter
io
r
i
s
eg
m
e
n
tal
SNR
an
d
o
v
er
s
u
b
tr
ac
tio
n
f
ac
to
r
ca
n
b
e
ca
lcu
late
f
r
o
m
E
q
u
at
io
n
(
8
)
[
5
]
.
e
l
s
e
D
D
Y
f
D
Y
S
i
2
2
2
2
2
2
)
(
ˆ
1
)
(
ˆ
)
(
)
(
ˆ
)
(
)
(
ˆ
;
m
a
x
0
m
i
n
0
)
(
S
N
R
S
N
R
(7
)
w
h
er
e
1
m
i
n
,
5
m
a
x
,
5
m
i
n
S
N
R
,
dB
20
m
a
x
S
NR
,
)
4
(
,
0
0
,
at
S
NR
dB
0
.
I
n
th
i
s
tech
n
iq
u
e
w
i
ll
u
n
if
o
r
m
t
h
e
n
o
is
e
e
f
f
ec
t
s
s
p
ec
tr
u
m
to
t
h
e
s
p
ee
ch
a
n
d
p
r
ed
ict
th
e
s
u
b
tr
ac
ti
o
n
f
ac
to
r
s
th
a
t
w
a
s
s
u
b
tr
ac
ted
n
o
is
y
b
y
o
v
er
-
e
s
ti
m
ate
o
f
n
o
is
e
f
ac
to
r
s
p
ec
tr
u
m
.
Sp
ee
ch
d
is
to
r
tio
n
a
n
d
r
e
m
n
a
n
t
m
u
s
ical
n
o
is
e
i
s
b
alan
ce
d
u
s
i
n
g
th
e
v
ar
io
u
s
co
m
b
in
at
io
n
s
o
f
o
v
er
-
s
u
b
tr
ac
tio
n
f
ac
to
r
an
d
s
p
ec
tr
al
f
lo
o
r
p
ar
a
m
eter
.
T
h
is
p
ar
am
eter
is
to
av
o
id
th
e
tr
ad
e
o
f
th
e
a
m
o
u
n
t
o
f
r
e
m
n
a
n
t
n
o
is
e
an
d
th
e
lev
el
o
f
p
er
ce
iv
e
d
m
u
s
ica
l
n
o
is
e.
I
f
p
ar
am
eter
v
al
u
e
is
lar
g
e,
it
p
r
o
d
u
ce
s
au
d
itab
le
n
o
is
e
d
u
e
to
a
v
er
y
litt
le
a
m
o
u
n
t
o
f
r
e
m
n
a
n
t
m
u
s
ical
n
o
is
e.
I
f
p
ar
a
m
e
ter
v
a
lu
e
is
v
er
y
s
m
a
ll
th
e
n
t
h
e
r
e
m
n
a
n
t
n
o
i
s
e
g
r
ea
t
l
y
r
ed
u
ce
d
b
u
t
s
p
ee
ch
s
ig
n
al
i
s
q
u
it
a
n
n
o
y
i
n
g
b
y
th
e
m
u
s
ical
n
o
is
e.
T
h
u
s
,
t
h
e
s
u
i
tab
le
d
esi
g
n
o
f
i
ts
p
ar
a
m
e
ter
v
al
u
e
is
s
et
f
o
llo
w
i
n
g
t
h
e
E
q
u
atio
n
(
7
)
an
d
p
ar
am
eter
is
s
et
to
0
.
0
3
.
A
s
s
u
ch
,
t
h
e
alg
o
r
it
h
m
ca
n
r
ed
u
ce
th
e
lev
el
o
f
p
er
ce
iv
ed
r
e
m
n
a
n
t
m
u
s
ical
n
o
is
e
w
h
ile
t
h
e
r
e
m
ai
n
in
g
o
f
t
h
e
b
ac
k
g
r
o
u
n
d
n
o
is
e
i
s
p
r
esen
ted
an
d
d
is
to
r
ted
th
e
e
n
h
an
ce
d
s
p
ee
ch
s
i
g
n
al.
Ma
n
y
t
y
p
e
o
f
r
esear
c
h
al
s
o
r
ep
o
r
ted
u
s
i
n
g
o
th
er
d
o
m
ai
n
,
e.
g
.
s
i
g
n
al
s
u
b
s
p
ac
e
ap
p
r
o
ac
h
[
1
]
,
[
3
]
.
I
t
d
if
f
er
s
f
r
o
m
th
e
s
p
ec
tr
al
s
u
b
tr
ac
tio
n
b
y
d
ec
o
m
p
o
s
in
g
t
h
e
n
o
is
y
s
p
ee
c
h
w
it
h
Kah
u
n
e
n
-
L
o
ev
e
-
T
r
an
s
f
o
r
m
(
KL
T
)
in
to
s
u
b
s
p
ac
e
th
at
o
cc
u
p
ied
p
r
i
m
ar
il
y
b
y
t
h
e
clea
n
s
p
ee
ch
v
ec
to
r
s
p
ac
e
s
ig
n
a
l
a
n
d
n
o
i
s
e
v
ec
to
r
s
p
ac
e
s
ig
n
al.
T
h
is
m
et
h
o
d
u
s
ed
K
L
T
in
s
tead
o
f
FF
T
w
h
i
ch
is
p
r
o
p
o
s
ed
b
y
s
p
ec
tr
al
s
u
b
tr
ac
tio
n
.
I
t
is
th
e
n
esti
m
ated
t
h
e
s
i
g
n
a
l
th
e
s
ig
n
al
o
f
in
ter
est
a
n
d
n
o
is
e
s
u
b
s
p
ac
e
f
r
o
m
a
s
u
b
s
p
ac
e
o
f
t
h
e
n
o
i
s
y
E
u
clid
ea
n
s
p
ac
e
[
1
]
.
I
n
[
1
]
m
en
tio
n
ed
t
h
a
t
t
h
er
e
ar
e
s
ev
er
al
d
if
f
er
en
t
t
y
p
e
s
o
f
t
h
e
s
p
ec
tr
al
s
u
b
tr
ac
tio
n
a
lg
o
r
it
h
m
s
f
a
m
il
y
.
A
cc
o
r
d
in
g
l
y
,
t
h
is
s
p
ec
tr
al
s
u
b
tr
ac
tio
n
t
y
p
e
esti
m
ate
s
t
h
e
s
p
ee
ch
b
y
s
u
b
tr
ac
ti
n
g
n
o
is
e
e
s
ti
m
atio
n
f
r
o
m
t
h
e
n
o
is
e
s
p
ee
ch
o
r
b
y
m
u
ltip
l
y
in
g
t
h
e
n
o
is
e
s
p
ec
tr
u
m
w
i
th
g
ain
f
u
n
ctio
n
s
,
an
d
th
en
co
m
b
i
n
e
it
w
i
th
t
h
e
p
h
ase
o
f
n
o
is
y
s
p
ee
ch
.
So
m
e
o
f
its
ex
a
m
p
le
s
,
in
b
r
ief
l
y
,
ar
e
s
p
ec
tr
al
o
v
er
-
s
u
b
t
r
ac
tio
n
,
s
p
ec
tr
al
s
u
b
tr
ac
tio
n
b
ased
o
n
p
er
ce
p
tu
al
p
r
o
p
er
ties
,
iter
ativ
e
s
p
ec
tr
al
s
u
b
tr
ac
tio
n
,
m
u
lti
-
b
a
n
d
s
p
ec
tr
al
s
u
b
tr
ac
tio
n
,
W
ie
n
er
f
ilter
in
g
.
T
h
er
ef
o
r
e,
s
p
ec
tr
al
s
u
b
tr
ac
ti
o
n
t
y
p
e
s
es
s
e
n
tiall
y
w
er
e
b
ased
o
n
in
t
u
it
iv
e
a
n
d
h
e
u
r
is
tical
l
y
b
ased
p
r
in
cip
les.
I
n
W
ien
er
f
ilter
t
y
p
e
al
g
o
r
ith
m
s
,
t
h
e
g
e
n
er
al
id
ea
is
to
m
in
i
m
ize
th
e
m
ea
n
s
q
u
ar
e
er
r
o
r
cr
iter
io
n
an
d
to
ac
h
iev
e
th
e
o
p
ti
m
al
f
ilter
as
m
en
tio
n
in
[1
]
,
[
1
8
]
.
T
h
e
ty
p
ical
f
o
r
m
u
la
o
f
t
h
e
W
ien
er
f
i
lter
w
it
h
n
o
n
ca
u
s
al
W
ien
er
f
ilter
f
o
r
w
h
ic
h
th
e
f
r
e
q
u
en
c
y
r
e
s
p
o
n
s
e
[
1
]
,
[
1
8
]
an
d
its
f
o
r
m
u
lar
ca
n
b
e
ex
p
r
ess
ed
as f
o
llo
w
2
2
2
2
2
.
ˆ
)
(
Y
S
D
E
S
E
S
E
W
i
e
n
e
r
A
an
d
)
(
)
(
)
(
)
(
d
s
s
W
i
e
n
e
r
P
P
P
(
8
)
w
h
er
e
]
[
E
is
as
s
i
g
n
ed
as
s
ig
n
al
es
ti
m
ato
r
an
d
p
ar
a
m
eter
an
d
is
ass
i
g
n
ed
to
s
o
m
e
co
n
s
ta
n
t.
T
h
ese
co
n
s
ta
n
t
r
ef
er
r
ed
as
p
ar
a
m
etr
ic
W
ien
er
f
ilter
s
in
w
h
ic
h
to
o
b
tai
n
t
h
eir
c
h
ar
ac
ter
is
tic
f
o
r
s
p
ee
c
h
s
o
lu
tio
n
.
I
n
E
q
u
atio
n
(
8
)
ass
ig
n
t
h
e
an
d
ar
e
eq
u
al
to
o
n
e.
T
h
u
s
,
th
e
e
n
h
an
ce
d
s
p
ee
ch
esti
m
atio
n
d
ep
en
d
s
lar
g
el
y
o
n
t
h
e
g
ai
n
p
ar
a
m
eter
’
s
i
m
p
r
o
v
e
m
e
n
t.
T
h
e
en
h
a
n
ce
d
s
p
ee
ch
est
i
m
at
io
n
a
n
d
its
g
ai
n
f
u
n
ctio
n
is
s
h
o
w
n
in
E
q
u
atio
n
(
9
)
.
T
h
is
g
ai
n
f
u
n
ctio
n
is
lar
g
el
y
d
ep
en
d
o
n
th
e
p
o
w
er
s
p
ec
tr
u
m
d
en
s
i
t
y
o
f
th
e
n
o
i
s
e
at
a
ce
r
tain
f
r
eq
u
e
n
c
y
th
at
at
ten
u
at
es e
ac
h
f
r
eq
u
en
c
y
co
m
p
o
n
e
n
t.
I
n
[
1
]
,
[
1
1
]
r
ev
ie
w
ed
th
e
s
tati
s
tical
m
o
d
el
b
ased
alg
o
r
it
h
m
.
I
ts
m
et
h
o
d
is
j
u
s
t
if
ied
b
y
t
h
e
s
t
atis
tics
o
f
sp
ee
ch
an
d
n
o
i
s
e
th
at
ar
e
n
o
t
av
ailab
le
an
d
th
er
e
i
s
n
o
k
n
o
w
led
g
e
o
f
t
h
e
b
est
d
is
to
r
ti
o
n
m
ea
s
u
r
e
in
t
h
e
p
er
ce
p
tio
n
s
en
s
e
b
y
m
o
d
if
ica
ti
o
n
o
f
u
s
i
n
g
Hid
d
en
Ma
r
k
o
v
M
o
d
el
(
HM
M)
b
ased
en
h
a
n
ce
m
en
t [
1
]
.
I
n
g
en
er
al,
th
is
m
et
h
o
d
ad
ap
ted
a
co
m
p
o
s
ite
s
o
u
r
ce
m
o
d
el
b
y
c
h
o
o
s
in
g
a
f
in
i
te
s
et
o
f
s
tat
is
tical
l
y
i
n
d
ep
en
d
en
t
Gau
s
s
ia
n
s
u
b
s
o
u
r
ce
s
m
o
d
el.
T
h
is
f
i
n
ite
s
et
i
s
co
n
s
id
er
as
s
w
i
tch
th
a
t
co
n
tr
o
lled
b
y
a
Ma
r
k
o
v
c
h
ai
n
.
T
h
e
HM
M
-
b
ased
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
S
in
g
le
C
h
a
n
n
el
S
p
ee
c
h
E
n
h
a
n
ce
men
t u
s
in
g
W
ien
er F
ilte
r
a
n
d
C
o
mp
r
ess
ive
S
en
s
in
g
(
A
ma
r
t S
u
lo
n
g
)
1945
en
h
a
n
ce
m
en
t
s
y
s
te
m
s
a
llo
w
a
s
ep
ar
atio
n
b
et
w
ee
n
s
p
ee
c
h
an
d
n
o
is
e
b
esid
e
t
h
at
it
i
n
t
r
o
d
u
ce
d
o
f
a
p
r
io
r
in
f
o
r
m
atio
n
ab
o
u
t
s
p
ee
ch
a
n
d
a
m
o
d
elin
g
o
f
n
o
is
e
lead
to
a
n
i
m
p
r
o
v
e
m
en
t
o
v
er
clas
s
ical
m
et
h
o
d
s
,
esp
ec
iall
y
at
lo
w
SN
R
s
a
n
d
f
o
r
s
p
ee
ch
co
r
r
u
p
ted
b
y
n
o
n
s
tatio
n
ar
y
n
o
is
e.
T
h
e
li
m
ita
tio
n
o
s
th
e
H
MM
-
b
ased
s
y
s
te
m
r
eq
u
ir
e
a
tr
ain
in
g
p
h
a
s
e
to
o
b
tain
th
e
s
p
e
ec
h
a
n
d
n
o
is
e
m
o
d
el
s
.
I
t
r
elativ
el
y
i
n
cr
ea
s
e
th
e
co
m
p
u
tatio
n
al
r
eq
u
ir
e
m
en
t.
T
h
e
ev
al
u
atio
n
o
f
th
i
s
s
tag
e
f
o
llo
w
ed
clea
n
s
p
ee
ch
est
i
m
at
io
n
u
s
i
n
g
Ma
x
i
m
u
m
A
P
o
s
ter
io
r
i
esti
m
atio
n
(
M
A
P
)
[
1
]
,
[
1
1
]
,
Min
i
m
u
m
Me
an
-
Sq
u
ar
e
E
r
r
o
r
(
MM
SE)
esti
m
atio
n
(
it
also
k
n
o
w
n
as
E
p
h
ar
ai
m
an
d
Ma
lla
h
’
s
e
s
ti
m
ato
r
)
[
1
9
]
.
T
h
is
m
et
h
o
d
[
1
9
]
f
o
cu
s
ed
o
n
p
r
o
d
u
cin
g
co
lo
r
les
s
r
esid
u
al
n
o
is
e
b
y
i
n
tr
o
d
u
ci
n
g
th
e
g
ai
n
f
u
n
ctio
n
’
s
esti
m
ato
r
a
s
a
f
u
n
ctio
n
o
f
a
p
o
s
teri
o
r
i S
N
R
an
d
a
p
r
io
r
i S
N
R
.
2
2
)
(
)
(
)
(
ˆ
Y
S
W
i
e
n
e
r
an
d
2
2
2
)
(
D
E
S
E
S
E
W
i
e
ne
r
(9
)
L
ater
,
[
1
8
]
p
r
o
p
o
s
ed
th
e
m
o
d
if
icatio
n
o
f
a
p
r
io
r
i
s
i
g
n
al
to
n
o
is
e
esti
m
atio
n
t
h
at
lead
s
t
o
th
e
b
es
t
s
u
b
j
ec
tiv
e
r
es
u
lt
s
an
d
ac
h
ie
v
e
d
th
e
tr
ad
e
-
o
f
f
b
et
w
ee
n
n
o
is
e
r
ed
u
ctio
n
s
w
it
h
lo
w
co
m
p
u
ta
tio
n
al
lo
ad
f
o
r
r
ea
l
ti
m
e
o
p
er
atio
n
s
.
Mo
r
eo
v
er
,
[
1
]
ad
ap
ted
w
it
h
a
n
o
n
-
ca
u
s
a
l
esti
m
ato
r
f
o
r
a
p
r
io
r
i
SN
R
a
n
d
a
co
r
r
esp
o
n
d
in
g
n
o
n
-
ca
u
s
al
to
e
n
h
a
n
ce
s
p
ee
ch
s
ig
n
al.
T
h
is
esti
m
ato
r
tec
h
n
i
q
u
e
p
r
o
d
u
ce
d
a
h
ig
h
er
i
m
p
r
o
v
e
m
en
t
i
n
s
e
g
m
en
tal
SNR
,
lo
w
er
lo
g
-
s
p
ec
tr
al
d
i
s
to
r
tio
n
,
an
d
b
etter
p
er
ce
p
tu
al
ev
a
lu
atio
n
o
f
s
p
ee
c
h
q
u
ali
t
y
a
s
s
e
s
s
m
e
n
t te
s
ts
(
P
E
SQ
s
co
r
es
b
ased
o
n
I
T
U
-
T
P
.
8
6
2
s
tan
d
ar
d
[
1
0
]
)
.
B
esid
es
th
at,
o
th
er
s
p
ee
ch
en
h
an
ce
m
e
n
t
tech
n
iq
u
e
s
[
1
]
,
[
4
]
also
in
tr
o
d
u
ce
d
.
I
n
[
1
6
]
m
e
n
tio
n
ed
th
e
m
o
d
i
f
icatio
n
o
f
b
o
o
s
ti
n
g
t
ec
h
n
iq
u
es a
n
d
its
ad
ap
tatio
n
t
o
te
m
p
o
r
al
m
as
k
i
n
g
th
r
es
h
o
ld
o
f
t
h
e
h
u
m
a
n
a
u
d
it
o
r
y
s
y
s
te
m
.
T
h
is
m
a
s
k
i
n
g
t
h
r
esh
o
ld
d
ep
en
d
s
o
n
h
u
m
a
n
au
d
ito
r
y
s
y
s
te
m
t
h
at
t
y
p
icall
y
u
s
in
g
i
n
s
p
ee
ch
a
n
d
au
d
io
co
d
in
g
to
lo
w
er
t
h
e
b
itr
ate
r
eq
u
ir
e
m
en
t.T
h
e
g
ain
f
u
n
c
tio
n
w
as
d
ep
en
d
ed
o
n
th
e
g
lo
b
al
f
o
r
w
ar
d
m
a
s
k
i
n
g
th
r
es
h
o
ld
an
d
f
o
r
w
ar
d
m
as
k
in
g
t
h
r
es
h
o
ld
in
ea
ch
s
u
b
b
an
d
[
1
6
]
.
I
t
ac
ted
as
th
e
f
ilter
o
p
er
atio
n
th
at
ex
p
r
es
s
ed
in
ti
m
e
d
o
m
ai
n
in
o
r
d
er
to
ev
alu
a
te
t
h
e
n
o
is
e
e
f
f
ec
t
s
to
t
h
e
s
p
ee
ch
s
i
g
n
a
l
i
n
ea
ch
s
u
b
b
an
d
.
3.
P
RO
P
O
SE
D
SPE
E
CH
E
NH
ANCEM
E
NT
A
L
G
O
RI
T
H
M
I
n
th
i
s
s
ec
tio
n
,
Fi
g
u
r
e
1
s
h
o
w
s
i
n
th
e
b
lo
ck
d
ia
g
r
a
m
o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
.
T
h
is
s
p
ee
c
h
en
h
a
n
ce
m
en
t a
lg
o
r
it
h
m
i
s
d
esi
g
n
ed
b
ased
o
n
W
ien
er
f
ilter
an
d
co
m
p
r
ess
i
v
e
s
e
n
s
i
n
g
(
C
S).
3
.
1
.
No
is
y
Sp
ec
t
ru
m
a
nd
U
pd
a
t
e
o
f
No
is
e
E
s
t
i
m
a
t
e
As
s
h
o
w
n
in
Fi
g
u
r
e
1
,
t
h
e
s
p
ee
ch
s
i
g
n
al
h
as
b
ee
n
co
n
ta
m
i
n
ated
b
y
n
o
is
e
a
n
d
it
i
s
w
ell
-
k
n
o
w
n
a
s
n
o
is
y
s
p
ee
ch
.
W
ith
t
h
i
s
m
et
h
o
d
,
th
e
n
o
is
y
s
p
ee
c
h
is
s
ep
ar
a
te
in
to
a
f
r
a
m
e
o
f
2
0
m
illi
s
ec
o
n
d
s
in
w
h
ich
ea
c
h
f
r
a
m
e
i
s
co
r
r
esp
o
n
d
ed
to
1
6
0
s
a
m
p
le
p
er
f
r
a
m
e
b
y
u
s
i
n
g
th
e
s
a
m
p
li
n
g
r
ate
o
f
8
k
Hz.
L
e
t
n
o
is
y
s
p
ee
ch
)
(
n
y
as
th
e
in
p
u
t
s
i
g
n
al
i
n
ter
m
o
f
ti
m
e
d
o
m
a
in
t
h
at
co
n
s
i
s
t
o
f
t
h
e
clea
n
s
p
ee
ch
)
(
n
s
an
d
ad
d
itiv
e
n
o
is
e
)
(
n
d
o
f
in
d
ep
en
d
en
t
s
o
u
r
ce
r
esp
ec
tiv
e
l
y
.
T
h
e
eq
u
atio
n
s
ar
e
r
estated
an
d
s
i
m
p
li
f
ied
in
o
r
d
er
to
m
a
k
e
u
n
d
er
s
tan
d
ab
le.
Fro
m
E
q
u
atio
n
(
1
)
an
d
(
2
)
,
n
o
is
e
esti
m
ate
[
2
0
]
w
i
th
th
e
h
y
p
o
th
e
s
is
f
o
r
m
u
la
ca
n
b
e
ex
p
r
ess
ed
i
n
E
q
u
atio
n
(
1
0
)
.
T
h
e
n
o
is
e
es
ti
m
atio
n
w
ill
ca
lc
u
late
b
ased
o
n
f
r
a
m
e
-
by
-
f
r
a
m
e
n
o
is
e
es
ti
m
atio
n
o
f
E
q
u
atio
n
(
1
1
)
.
T
h
e
h
y
p
o
t
h
esi
s
o
f
E
q
u
at
io
n
(
1
0
)
is
u
p
d
ate
th
e
n
o
is
e
es
ti
m
ate
2
D
.
T
h
e
r
an
g
1
0
d
d
w
a
s
ass
i
g
n
ed
f
o
r
s
m
o
o
t
h
in
g
f
ac
t
o
r
.
)
(
0
H
an
d
)
(
1
H
d
en
o
ted
th
e
s
p
ee
ch
ab
s
en
t
an
d
t
h
e
s
p
ee
c
h
p
r
esen
t
h
y
p
o
t
h
esi
s
r
esp
ec
tiv
e
l
y
.
He
n
ce
,
th
e
n
o
is
e
e
s
ti
m
ate
)
(
D
o
b
tain
ed
f
r
o
m
E
q
u
a
tio
n
(
1
1
)
w
h
er
e
k
Y
k
H
P
k
p
,
|
,
)
,
(
'
'
1
d
en
o
ted
as
t
h
e
s
p
ee
c
h
p
r
esen
ce
p
r
o
b
ab
ilit
y
o
f
t
h
e
n
o
is
e
v
ar
ia
n
ce
t
h
at
co
r
r
u
p
ted
in
h
ig
h
n
o
n
s
ta
tio
n
ar
y
n
o
is
e
e
n
v
ir
o
n
m
en
t
s
.
2
2
)
(
D
E
D
an
d
)
(
ˆ
)
1
(
ˆ
:
)
(
1
)
(
ˆ
)
1
(
ˆ
:
)
(
2
2
'
1
2
2
2
'
0
D
D
d
D
d
D
H
Y
H
(
1
0
)
2
)
(
))
(
1
(
)
1
(
)
(
)
(
Y
D
D
s
s
an
d
)
(
'
)
1
(
)
(
p
d
d
s
(
1
1
)
3
.
2
.
SNR
E
s
t
i
m
a
t
o
r
a
nd
Wiener
F
ilte
r
T
h
e
SNR
esti
m
ato
r
is
r
ep
r
esen
ted
b
y
o
b
s
er
v
i
n
g
lo
ca
l
a
p
o
s
teri
o
r
i
S
N
R
an
d
a
p
r
io
r
i
S
N
R
i
n
E
q
u
atio
n
(
1
2
)
r
esp
ec
tiv
el
y
.
T
h
is
es
ti
m
a
to
r
w
as
ad
ap
ted
b
y
u
s
i
n
g
[
1
9
]
in
o
r
d
er
to
p
r
o
d
u
ce
co
lo
r
less
r
esid
u
al
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.
7
,
No
.
4
,
A
u
g
u
s
t
2017
:
1
9
4
1
–
1
9
5
1
1946
n
o
is
e
an
d
to
i
m
p
r
o
v
e
t
h
e
g
ai
n
f
u
n
ctio
n
o
f
th
e
W
ie
n
er
f
ilter
.
T
h
e
)
(
ˆ
1
n
S
is
th
e
p
r
ev
io
u
s
f
r
a
m
e
e
s
ti
m
atio
n
s
p
ee
ch
,
w
h
er
e
1
p
o
s
t
S
N
R
is
i
n
ter
p
r
eted
as
in
s
tan
tan
eo
u
s
S
N
R
(
i
n
s
t
S
N
R
)
w
h
ile
98
.
0
an
d
y
y
P
)
(
0
y
if
an
d
0
)
(
y
P
o
th
erw
is
e
.
T
h
is
W
ien
er
tech
n
iq
u
e
in
E
q
u
atio
n
(
1
3
)
w
a
s
m
o
d
if
ied
b
ased
o
n
[
1
8
]
t
o
o
b
tain
th
e
h
i
g
h
a
m
p
l
itu
d
e
s
p
e
ctr
u
m
w
ei
g
h
t
esti
m
ate
w
h
e
n
ap
p
ly
i
n
g
E
q
u
atio
n
(
1
2
)
to
n
o
n
-
lin
ea
r
o
p
ti
m
al
g
ai
n
f
u
n
ctio
n
o
f
E
q
u
atio
n
(
1
3
)
an
d
p
r
o
d
u
ce
d
th
e
en
h
a
n
ce
d
s
p
ee
ch
s
i
g
n
al.
T
h
is
m
o
d
i
f
ied
tech
n
i
q
u
e
w
il
l
r
ed
u
ce
th
e
m
is
m
atc
h
w
e
ig
h
t
o
f
t
h
e
i
n
ter
este
d
s
i
g
n
a
l.
T
h
en
,
t
h
e
i
n
v
er
s
e
FF
T
tr
an
s
f
o
r
m
ed
is
s
y
n
th
e
s
is
.
I
t
also
d
er
iv
ed
u
n
d
er
as
s
u
m
p
tio
n
t
h
at
o
f
a
k
e
y
p
ar
a
m
eter
i
n
th
e
r
ed
u
ctio
n
o
f
t
h
e
n
o
i
s
e
a
n
d
i
m
p
r
o
v
i
n
g
t
h
e
s
p
ee
ch
d
i
s
to
r
tio
n
w
h
er
e
th
e
tec
h
n
iq
u
e
g
iv
e
n
a
d
ec
is
io
n
-
d
ir
ec
ted
m
et
h
o
d
as lo
w
co
m
p
u
tatio
n
al
l
o
ad
f
o
r
r
ea
l ti
m
e
o
p
er
atio
n
.
2
2
)
(
)
(
D
Y
S
N
R
p
o
s
t
an
d
)
1
(
1
)
(
)
(
ˆ
)
(
)
(
2
2
1
2
2
p
o
s
t
n
p
r
i
o
S
N
R
P
D
S
D
S
E
S
N
R
(
1
2
)
X
j
W
i
e
n
e
r
e
Y
I
F
F
T
n
y
ˆ
an
d
p
r
i
o
p
r
i
o
W
i
e
n
e
r
S
N
R
S
N
R
1
)
(
(
1
3
)
N
o
i
s
e
A
v
e
r
a
g
e
N
o
i
s
y
S
p
e
e
c
h
E
s
t
i
m
a
t
o
r
S
p
e
c
t
r
a
l
S
N
R
E
s
t
i
m
a
t
o
r
W
i
e
n
e
r
F
i
l
t
e
r
C
o
m
p
r
e
s
s
i
v
e
S
e
n
s
i
n
g
(
C
S
)
M
o
d
i
f
i
c
a
t
i
o
n
N
o
i
s
y
S
p
e
e
c
h
E
n
h
a
n
c
e
d
S
p
e
e
c
h
N
o
i
s
e
E
s
t
i
m
a
t
e
N
o
i
s
y
S
p
e
c
t
r
u
m
S
N
R
E
s
t
i
m
a
t
o
r
W
i
e
n
e
r
F
i
l
t
e
r
C
o
m
p
r
e
s
s
i
v
e
S
e
n
s
i
n
g
(
C
S
)
M
o
d
i
f
i
c
a
t
i
o
n
N
o
i
s
y
S
p
e
e
c
h
E
n
h
a
n
c
e
d
S
p
e
e
c
h
P
E
S
Q
M
e
a
s
u
r
e
P
E
S
Q
S
c
o
r
e
C
l
e
a
n
S
p
e
e
c
h
Fig
u
r
e
1
.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
it
h
m
b
ased
o
n
W
ie
n
er
f
ilter
a
n
d
co
m
p
r
es
s
iv
e
s
en
s
i
n
g
tec
h
n
iq
u
e
3
.
3
.
Co
m
press
iv
e
Sens
i
ng
M
o
difi
ca
t
io
n
T
h
e
co
m
p
r
ess
i
v
e
s
e
n
s
i
n
g
(
C
S)
tech
n
iq
u
e
is
also
m
o
d
i
f
ied
.
T
h
is
n
o
v
el
C
S
ap
p
r
o
ac
h
is
f
u
n
d
a
m
en
tall
y
d
if
f
er
e
n
t
f
r
o
m
t
h
e
w
ell
-
k
n
o
wn
Sh
a
n
n
o
n
s
a
m
p
li
n
g
t
h
eo
r
e
m
[
6
]
.
T
h
is
tech
n
iq
u
e
u
s
ed
s
a
m
p
lin
g
t
h
eo
r
y
th
at
o
f
s
elec
ti
n
g
th
e
i
n
ter
es
ted
s
i
g
n
al
a
n
d
r
ec
o
v
er
w
it
h
al
m
o
s
t
ex
ac
t
s
i
g
n
al
r
ec
o
n
s
tr
u
cti
o
n
f
r
o
m
n
o
i
s
eles
s
o
b
s
er
v
atio
n
s
[
6
]
,
[
9
]
.
T
h
e
m
aj
o
r
ad
v
an
tag
e
o
f
th
e
C
S
is
th
e
r
ec
o
v
er
y
p
r
ed
ictio
n
s
o
f
th
e
s
i
g
n
a
ls
f
r
o
m
in
co
m
p
lete
m
ea
s
u
r
e
m
e
n
ts
(
in
f
o
r
m
at
io
n
)
t
h
at
w
a
s
ap
p
lied
i
n
v
ar
io
u
s
ap
p
licatio
n
s
.
Mo
r
eo
v
er
,
th
e
C
S tec
h
n
iq
u
e
r
elies
o
n
t
h
e
k
e
y
e
f
f
icien
c
y
o
f
th
e
e
m
p
ir
ical
o
b
s
er
v
atio
n
w
i
t
h
w
el
l
s
p
ar
s
e
ap
p
r
o
x
i
m
at
io
n
s
in
s
u
itab
le
b
asi
s
b
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
S
in
g
le
C
h
a
n
n
el
S
p
ee
c
h
E
n
h
a
n
ce
men
t u
s
in
g
W
ien
er F
ilte
r
a
n
d
C
o
mp
r
ess
ive
S
en
s
in
g
(
A
ma
r
t S
u
lo
n
g
)
1947
o
n
l
y
a
s
m
all
a
m
o
u
n
t
o
f
n
o
n
z
er
o
co
ef
f
icie
n
ts
[
6
]
,
[
9
]
.
T
h
e
C
S
m
e
th
o
d
u
s
ed
g
r
ad
ien
t
p
r
o
j
ec
tio
n
f
o
r
s
p
ar
s
e
r
ec
o
n
s
tr
u
ctio
n
(
GP
SR
)
to
ex
p
er
i
m
en
tall
y
in
v
es
tig
a
te
th
e
i
n
t
er
ac
tiv
e
ef
f
ec
t
s
o
f
t
h
e
co
r
r
u
p
t
ed
n
o
is
e
an
d
o
b
tain
b
etter
i
m
p
r
o
v
e
m
e
n
t
to
th
e
li
s
t
en
er
w
ith
n
o
is
ele
s
s
r
ed
u
ctio
n
[
2
1
]
.
T
h
is
m
et
h
o
d
ap
p
lied
b
ased
o
n
th
e
w
ei
g
h
t
ad
ap
tatio
n
(
w
Ax
n
y
)
(
ˆ
)
o
f
in
v
er
s
e
f
ast
Fo
u
r
ier
tr
an
s
f
o
r
m
in
E
q
u
atio
n
(
1
4
)
to
ac
h
iev
e
h
i
g
h
q
u
alit
y
n
o
is
e
r
ed
u
ct
io
n
a
n
d
en
h
a
n
c
e
s
p
ee
ch
s
i
g
n
al
Ax
n
s
)
(
ˆ
w
h
er
e
t
h
e
n
at
u
r
e
o
f
a
m
atr
ic
is
d
ef
i
n
ed
b
y
m
ea
s
u
r
e
m
e
n
t
m
atr
ix
n
m
R
A
.
T
h
e
est
i
m
ated
co
e
f
f
icien
t
n
R
x
,
an
d
m
o
d
el
m
is
m
atc
h
m
R
w
is
u
n
d
er
ass
u
m
p
tio
n
t
h
at
n
m
.
T
o
r
ec
o
v
er
th
e
ill
-
p
o
s
ed
co
n
d
itio
n
o
f
s
i
g
n
a
l
w
it
h
s
u
f
f
icie
n
t
s
p
ar
s
e
x
o
f
u
n
co
n
s
tr
ai
n
ed
p
r
o
b
lem
u
s
ed
th
e
GP
SR
[
3
6
]
tech
n
iq
u
e,
w
h
e
r
e
th
e
s
p
u
r
io
u
s
co
m
p
o
n
en
ts
m
R
w
a
r
e
r
ed
u
ce
d
n
o
is
eles
s
d
is
to
r
tio
n
s
.
T
h
is
tech
n
iq
u
e
ca
n
b
e
ex
p
r
ess
ed
as
in
E
q
u
atio
n
(
1
4
).
1
2
2
2
1
m
i
n
x
Ax
y
x
(
1
4
)
L
et
t
h
e
s
a
m
p
le
y
is
in
p
u
t
w
ei
g
h
s
i
g
n
al
co
r
r
elatio
n
to
p
r
ed
eter
m
in
ed
t
h
e
ele
m
e
n
t
o
f
w
ei
g
h
ad
ap
tatio
n
)
(
ˆ
n
y
.
T
h
e
d
eter
m
in
a
tio
n
to
ex
ac
t
s
o
lu
tio
n
o
f
t
h
e
s
p
ar
s
e
r
ec
o
v
er
y
y
is
u
til
ized
to
r
eg
u
late
th
e
r
ec
o
v
er
y
o
f
t
h
e
esti
m
ated
co
ef
f
icien
t
i
n
t
h
e
p
r
ed
icted
s
ig
n
al
x
ˆ
o
f
x
an
d
ac
h
i
ev
e
th
e
i
m
p
r
o
v
e
m
en
t
o
f
s
p
ee
ch
q
u
alit
y
w
it
h
n
o
is
e
r
ed
u
ctio
n
.
T
h
is
C
S
m
o
d
if
icatio
n
tec
h
n
iq
u
e
r
elies
o
n
th
e
k
e
y
e
f
f
icien
c
y
o
f
t
h
e
em
p
ir
ical
o
b
s
er
v
atio
n
w
it
h
w
ell
s
p
ar
s
e
ap
p
r
o
x
i
m
a
tio
n
in
s
u
i
tab
le
b
asis
b
y
o
n
l
y
s
m
a
ll a
m
o
u
n
t o
f
n
o
n
ze
r
o
co
ef
f
icie
n
ts
[
6
]
,
[
9
]
.
4.
E
XP
E
R
I
M
E
NT
A
L
RE
SUL
T
S AN
D
D
I
SC
USS
I
O
NS
P
E
SQ
o
b
j
ec
tiv
e
ass
es
s
m
e
n
t
t
est
a
n
d
its
p
er
ce
n
ta
g
e
i
m
p
r
o
v
e
m
en
t
i
n
w
as
in
v
es
tig
a
te
i
n
w
h
ic
h
to
ev
alu
a
te
th
e
e
n
h
an
ce
m
e
n
t
o
f
th
e
s
p
ee
c
h
s
i
g
n
al
an
d
t
h
e
n
co
m
p
ar
e
w
it
h
t
h
e
clea
n
s
p
ee
ch
s
i
g
n
al
t
h
at
o
f
a
p
ar
ticu
lar
ass
es
s
m
en
t
s
ig
n
al
[
1
]
,
[
24
]
,
[
2
9
]
.
T
h
e
P
E
SQ
s
co
r
e
h
as
al
m
o
s
t
co
r
r
elate
d
w
it
h
s
u
b
j
ec
tiv
e
ass
es
s
m
e
n
t
test
o
f
a
9
3
.
5
%
co
r
r
elatio
n
w
h
ile
o
th
er
o
b
j
ec
tiv
e
test
s
u
ch
as
I
tak
u
r
a
-
s
a
ito
d
is
to
r
tio
n
alg
o
r
ith
m
,
A
r
tic
u
latio
n
in
d
ex
,
s
e
g
m
e
n
t
SN
R
,
an
d
SN
R
h
av
e
co
r
r
elatio
n
ass
e
s
s
m
e
n
t
test
o
f
5
9
%,
6
7
%,
7
7
%,
an
d
2
4
%
r
esp
ec
tiv
el
y
[
1
6
]
.
I
n
[
1
6
]
also
in
tr
o
d
u
ce
d
th
e
n
e
w
s
p
ee
ch
q
u
alit
y
ass
es
s
m
en
t
test
i
n
ter
m
o
f
p
er
ce
n
tag
e
P
E
SQ
i
m
p
r
o
v
e
m
en
t
.
T
h
is
p
er
ce
n
tag
e
i
m
p
r
o
v
e
m
e
n
t c
a
n
b
e
ex
p
r
ess
ed
as sh
o
w
n
i
n
E
q
u
atio
n
(
1
5
).
%
100
r
e
f
r
e
f
p
r
o
c
P
E
S
Q
P
E
S
Q
P
E
S
Q
(
15
)
E
q
u
atio
n
(
1
5
)
m
e
n
tio
n
ed
o
n
p
r
o
c
P
E
S
Q
a
n
d
r
e
f
P
E
S
Q
,
it d
en
o
ted
th
e
o
b
j
ec
tiv
e
P
E
SQ a
s
s
ess
m
e
n
t
s
co
r
e
o
f
th
e
en
h
a
n
ce
d
s
p
ee
ch
co
m
p
ar
ed
w
ith
t
h
e
clea
n
s
p
ee
c
h
s
i
g
n
al
wh
ile
i
n
r
e
f
P
E
S
Q
r
ef
er
s
to
P
E
SQ
s
co
r
e
o
f
test
i
n
g
n
o
is
y
s
p
ee
ch
p
er
f
o
r
m
a
n
ce
q
u
a
lit
y
co
m
p
ar
ed
w
it
h
th
e
clea
n
s
p
ee
ch
r
esp
ec
tiv
el
y
.
T
h
e
f
o
u
r
d
i
f
f
er
e
n
t
r
ea
l
ar
ti
f
ic
ial
ad
d
ed
f
o
r
m
t
h
e
n
o
is
y
s
p
e
ec
h
co
r
p
u
s
(
NOI
Z
E
U
S)
I
E
E
E
s
tan
d
ar
d
1
9
9
6
[
1
,
2
2
]
T
h
ese
n
o
is
y
d
at
a
s
et
u
s
ed
th
e
Am
er
ican
E
n
g
li
s
h
la
n
g
u
a
g
e,
w
h
er
e
t
h
e
s
p
ee
c
h
o
r
ig
in
all
y
s
a
m
p
led
at
2
5
k
Hz
an
d
d
o
w
n
-
s
a
m
p
led
to
8
k
Hz.
T
h
e
tr
ad
itio
n
al
alg
o
r
ith
m
s
i
n
cl
u
d
e
Sp
s
u
b
[
2
3
]
,
Ss
r
d
c
[
2
4
]
,
P
k
lt
[
2
5
]
,
W
n
r
W
t
[
2
6
]
,
Mm
a
s
k
[
2
7
]
,
an
d
m
m
s
e
[
1
9
]
r
esp
ec
tiv
el
y
.
T
h
e
P
E
SQ
ass
e
s
s
m
e
n
t
test
w
a
s
u
s
ed
to
e
v
al
u
ate
th
e
m
ai
n
a
n
al
y
s
is
an
d
i
ts
s
i
g
n
i
f
ic
an
t
d
i
f
er
en
t
b
et
w
ee
n
th
e
p
r
o
p
o
s
ed
Sp
E
n
C
S
an
d
th
e
o
t
h
er
al
g
o
r
ith
m
s
at
v
ar
io
u
s
n
o
is
e
t
y
p
e
SN
R
s
.
Fig
u
r
e
2
cl
ea
r
l
y
i
n
d
ic
ated
th
e
i
m
p
r
o
v
e
m
en
t
o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
in
t
h
e
w
a
v
e
f
o
r
m
s
an
d
s
p
ec
tr
o
g
r
a
m
r
esu
lts
w
h
e
n
co
m
p
ar
e
w
i
th
tr
ad
itio
n
al
al
g
o
r
ith
m
s
,
n
o
is
y
s
p
ee
c
h
.
I
n
Fi
g
u
r
e
3
,
th
e
P
E
SQ sco
r
e
in
t
h
e
p
r
o
p
o
s
ed
Sp
E
n
C
S
a
lg
o
r
ith
m
o
u
tp
er
f
o
r
m
s
th
e
s
p
ee
c
h
q
u
a
lit
y
co
m
p
ar
ed
to
o
v
er
all
s
co
r
e
w
it
h
o
t
h
er
alg
o
r
ith
m
s
o
f
all
n
o
i
s
e
t
y
p
es,
i
.
e.
0
,
5
,
1
0
,
1
5
d
B
SNR
.
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.
7
,
No
.
4
,
A
u
g
u
s
t
2017
:
1
9
4
1
–
1
9
5
1
1948
(
a)
T
h
e
s
p
ee
ch
w
a
v
e
f
o
r
m
o
f
cl
ea
n
s
p
ee
ch
(
b
)
T
h
e
s
p
ee
ch
w
a
v
e
f
o
r
m
o
f
c
lean
s
p
ee
ch
(
c)
T
h
e
s
p
ee
ch
w
a
v
e
f
o
r
m
o
f
t
h
e
p
r
o
p
o
s
ed
Sp
E
n
C
S
alg
o
r
ith
m
(
d
)
T
h
e
s
p
ee
ch
w
a
v
e
f
o
r
m
o
f
P
k
lt a
l
g
o
r
ith
m
(
e)
T
h
e
s
p
ee
ch
w
a
v
e
f
o
r
m
o
f
m
m
s
e
al
g
o
r
ith
m
(
f
)
T
h
e
s
p
ec
tr
o
g
r
am
o
f
clea
n
s
p
ee
ch
(
g
)
T
h
e
s
p
ec
tr
o
g
r
am
o
f
n
o
is
y
s
p
ee
ch
(
h
)
T
h
e
s
p
ec
tr
o
g
r
am
o
f
th
e
p
r
o
p
o
s
ed
Sp
E
n
C
S
alg
o
r
ith
m
(
i)
T
h
e
s
p
ec
tr
o
g
r
a
m
o
f
P
k
lt a
l
g
o
r
ith
m
(
j
)
T
h
e
s
p
ec
tr
o
g
r
a
m
o
f
m
m
s
e
alg
o
ith
m
Fig
u
r
e
2
.
T
h
e
co
m
p
ar
is
o
n
o
f
s
p
ee
ch
w
a
v
e
f
o
r
m
(
i.e
.
a
-
e)
an
d
its
s
p
ec
tr
o
g
r
a
m
(
i.e
.
f
-
j
)
o
f
th
e
p
r
o
p
o
s
ed
Sp
E
n
C
S a
lg
o
r
it
h
m
th
at
o
f
air
p
o
r
t n
o
is
e
at“sp
1
2
.
w
av
”
at
0
d
B
SNR
(
a
)
T
h
e
P
ESQ
a
sse
ssm
e
n
t
s
c
o
r
e
o
f
A
i
r
p
o
r
t
N
o
i
se
(
b
)
T
h
e
P
ESQ
a
sse
ssm
e
n
t
sco
r
e
o
f
B
a
b
b
l
e
N
o
i
se
(
c
)
T
h
e
P
ESQ
a
sse
ssm
e
n
t
s
c
o
r
e
o
f
C
a
r
N
o
i
se
(
d
)
T
h
e
P
ESQ
a
sse
ssm
e
n
t
sco
r
e
o
f
Ex
h
i
b
i
t
i
o
n
N
o
i
se
Fig
u
r
e
3
.
C
o
m
p
ar
is
o
n
o
f
P
E
S
Q
ass
es
s
m
en
t te
s
t o
f
t
h
e
p
r
o
p
o
s
ed
Sp
E
n
C
S
al
g
o
r
it
h
m
w
it
h
o
th
er
co
n
v
e
n
tio
n
a
l a
lg
o
r
i
th
m
s
at
0
d
B
,
5
d
B
,
1
0
d
B
an
d
1
5
d
B
r
esp
ec
tiv
el
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
S
in
g
le
C
h
a
n
n
el
S
p
ee
c
h
E
n
h
a
n
ce
men
t u
s
in
g
W
ien
er F
ilte
r
a
n
d
C
o
mp
r
ess
ive
S
en
s
in
g
(
A
ma
r
t S
u
lo
n
g
)
1949
T
ab
le
1
.
T
h
e
P
E
SQ im
p
r
o
v
e
m
en
t in
p
er
ce
n
tag
e
(
%)
o
f
t
h
e
p
r
o
p
o
s
ed
Sp
E
n
C
S c
o
m
p
ar
es
w
it
h
o
th
er
alg
o
r
it
h
m
s
T
ab
le
1
s
h
o
w
n
t
h
at
t
h
e
w
o
r
s
t
ca
s
e
ap
p
ea
r
w
it
h
0
d
B
at
all
ty
p
e
o
f
n
o
i
s
e
co
n
d
itio
n
s
.
Mo
s
t
o
f
P
E
SQ
p
er
ce
n
tag
e
i
m
p
r
o
v
e
m
e
n
t
r
es
u
lts
o
f
t
h
e
tr
ad
itio
n
al
al
g
o
r
ith
m
s
w
er
e
b
elo
w
1
0
%
an
d
its
i
m
p
r
o
v
e
m
e
n
t
r
e
m
ai
n
u
n
co
n
s
i
s
te
n
c
y
.
I
t
w
a
s
o
n
l
y
a
t
m
m
s
e
alg
o
r
it
h
m
p
r
o
d
u
ce
d
co
m
p
ar
ab
le
r
esu
l
ts
w
it
h
th
e
p
r
o
p
o
s
ed
Sp
E
n
C
S
alg
o
r
ith
m
.
T
h
e
o
v
er
all
av
er
a
g
e
o
f
t
h
e
i
m
p
r
o
v
e
m
e
n
t
i
n
th
e
p
r
o
p
o
s
ed
Sp
E
n
C
S
is
ar
o
u
n
d
2
0
%
to
all
n
o
is
y
ass
es
s
m
en
t te
s
ts
b
u
t o
th
er
alg
o
r
ith
m
p
r
o
d
u
ce
d
less
t
h
a
n
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
s
.
5.
CO
NCLU
SI
O
NS A
ND
F
UT
URE WO
RK
S
A
n
e
w
s
p
ee
c
h
en
h
a
n
ce
m
e
n
t
a
p
p
r
o
ac
h
b
y
u
s
i
n
g
W
ie
n
er
f
ilte
r
an
d
co
m
p
r
es
s
i
v
e
s
e
n
s
i
n
g
w
a
s
p
r
o
p
o
s
ed
f
o
r
en
h
an
ci
n
g
s
p
ee
c
h
d
eg
r
ad
ed
b
y
ad
d
iti
v
e
n
o
is
e.
T
h
e
n
o
i
s
e
esti
m
atio
n
i
s
ad
ap
ted
in
wh
ich
to
tr
ac
k
n
o
is
e
u
p
d
ate
esti
m
atio
n
co
n
ti
n
u
o
u
s
l
y
.
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
is
b
ased
o
n
t
h
e
W
ie
n
er
f
ilter
an
d
co
m
p
r
e
s
s
i
v
e
s
en
s
in
g
.
T
h
e
W
ien
er
f
ilter
is
m
o
d
i
f
ied
f
o
r
r
ed
u
cin
g
co
lo
r
less
r
esid
u
al
n
o
is
e
b
ef
o
r
e
W
ien
er
f
ilter
is
ca
lcu
lated
.
W
ien
er
f
il
ter
is
t
h
e
n
p
r
o
d
u
ce
d
th
e
o
p
ti
m
al
g
ai
n
w
it
h
i
n
cr
ea
s
in
g
a
m
p
lit
u
d
e
s
p
ec
tr
u
m
weig
h
t
esti
m
ate
an
d
r
ed
u
cin
g
m
is
m
atc
h
s
i
g
n
al
est
i
m
ate.
T
h
e
co
m
p
r
ess
i
v
e
s
e
n
s
in
g
later
is
m
o
d
if
ied
to
p
r
ed
ict
th
e
i
n
ter
ested
s
i
g
n
als
f
r
o
m
i
n
co
m
p
le
te
m
ea
s
u
r
e
m
e
n
ts
(
s
i
g
n
al
s
)
an
d
r
ec
o
v
er
w
i
th
al
m
o
s
t
s
i
g
n
a
l
r
ec
o
n
s
tr
u
cti
o
n
f
r
o
m
n
o
i
s
eles
s
o
b
s
er
v
atio
n
s
.
O
u
r
in
v
esti
g
at
io
n
an
d
ev
al
u
atio
n
o
f
t
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
s
o
u
tp
er
f
o
r
m
s
t
h
e
o
th
er
co
n
v
e
n
tio
n
al
al
g
o
r
it
h
m
s
at
v
ar
io
u
s
n
o
i
s
e
t
y
p
es.
ACK
NO
WL
E
D
G
E
M
E
NT
S
T
h
is
r
esear
ch
h
a
s
b
ee
n
s
u
p
p
o
r
ted
b
y
I
n
ter
n
a
tio
n
al
I
s
la
m
i
c
Un
i
v
er
s
i
t
y
Ma
la
y
s
ia
R
e
s
ea
r
ch
Gr
an
t,
R
I
GS1
6
-
336
-
0
500.
RE
F
E
R
E
NC
E
S
[1
]
P
.
C.
L
o
izo
u
,
“
S
p
e
e
c
h
En
h
a
n
c
e
m
e
n
t:
T
h
e
o
ry
a
n
d
P
ra
c
ti
c
e
,
”
CRC
P
re
ss
,
2
0
1
3
.
[2
]
R.
S
u
d
irg
a
,
“
A
S
p
e
e
c
h
En
h
a
n
c
e
m
e
n
t
S
y
ste
m
Ba
se
d
o
n
S
tatisti
c
a
l
a
n
d
A
c
o
u
stic
-
P
h
o
n
e
ti
c
K
n
o
w
led
g
e
,
”
2
0
0
9
.
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.
7
,
No
.
4
,
A
u
g
u
s
t
2017
:
1
9
4
1
–
1
9
5
1
1950
[3
]
N.
Up
a
d
h
y
a
y
,
A
.
Ka
r
m
a
k
a
r,
“
S
p
e
e
c
h
En
h
a
n
c
e
m
e
n
t
u
si
n
g
S
p
e
c
tral
S
u
b
trac
ti
o
n
-
ty
p
e
A
lg
o
rit
h
m
s:
A
Co
m
p
a
riso
n
a
n
d
S
im
u
latio
n
S
t
u
d
y
”
,
Pro
c
e
d
ia
C
o
mp
u
ter
S
c
ien
c
e
, v
o
l
.
5
4
,
p
p
.
5
7
4
_
5
8
4
,
2
0
1
5
.
[4
]
S
.
V
.
V
a
se
g
h
i,
“
A
d
v
a
n
c
e
d
Dig
it
a
l
S
ig
n
a
l
P
r
o
c
e
ss
in
g
a
n
d
n
o
ise
R
e
d
u
c
ti
o
n
,
”
Jo
h
n
W
il
e
y
&
S
o
n
s,
2
0
0
8
.
[5
]
N.
Up
a
d
h
y
a
y
,
A
.
Ka
r
m
a
k
a
r,
“
S
in
g
le
-
Ch
a
n
n
e
l
S
p
e
e
c
h
E
n
h
a
n
c
e
m
e
n
t
u
sin
g
Crit
ica
l
-
Ba
n
d
Ra
t
e
S
c
a
le
Ba
se
d
Im
p
ro
v
e
d
M
u
lt
i
-
Ba
n
d
S
p
e
c
tral
S
u
b
trac
ti
o
n
,
”
J
o
u
rn
a
l
o
f
S
i
g
n
a
l
a
n
d
In
f
o
rm
a
ti
o
n
Pro
c
e
ss
in
g
,
v
o
l.
4
,
n
o
.
3
,
pp.
3
1
4
-
3
2
6
,
Ju
l
.
2
0
1
3
.
[6
]
M
.
F
o
r
n
a
sie
r,
H.
Ra
u
h
u
t,
“
Co
m
p
r
e
ss
iv
e
se
n
sin
g
.
In
Ha
n
d
b
o
o
k
o
f
M
a
th
e
m
a
ti
c
a
l
M
e
th
o
d
s
in
I
m
a
g
in
g
”
,
p
p
.
1
8
7
-
2
2
8
,
S
p
rin
g
e
r
Ne
w
Yo
rk
,
2
0
1
1
.
[7
]
M
.
Un
se
r,
“
S
a
m
p
li
n
g
-
5
0
y
e
a
rs
a
f
ter
S
h
a
n
n
o
n
”
,
P
r
o
c
e
e
d
in
g
s o
f
th
e
IEE
E,
V
o
l.
8
8
,
n
o
.
4
,
p
p
.
5
6
9
_
5
8
7
,
A
p
r.
2
0
0
0
.
[8
]
M
.
Un
se
r,
“
S
a
m
p
li
n
g
-
5
0
y
e
a
rs
a
f
ter
S
h
a
n
n
o
n
”
,
P
r
o
c
e
e
d
in
g
s o
f
th
e
IEE
E,
V
o
l.
8
8
,
n
o
.
4
,
p
p
.
5
6
9
_
5
8
7
,
A
p
r.
2
0
0
0
.
[9
]
R.
G
.
Ba
ra
n
iu
k
,
“
Co
m
p
re
ss
iv
e
S
e
n
sin
g
”
,
IEE
E
S
i
g
n
a
l
Pr
o
c
e
ss
in
g
M
a
g
a
zin
e
,
v
o
l.
2
4
,
n
o
.
4
,
J
u
l.
2
0
0
7
.
[1
0
]
IT
U
-
T
,
R.
P
.
8
6
2
,
“
P
e
rc
e
p
tu
a
l
Ev
a
lu
a
ti
o
n
o
f
S
p
e
e
c
h
Q
u
a
li
ty
(P
E
S
Q):
A
n
Ob
jec
ti
v
e
M
e
th
o
d
f
o
r
e
n
d
-
to
-
e
n
d
S
p
e
e
c
h
Qu
a
li
ty
A
ss
e
ss
m
e
n
t
o
f
n
a
rro
w
-
b
a
n
d
T
e
lep
h
o
n
e
Ne
tw
o
rk
s
a
n
d
S
p
e
e
c
h
C
o
d
e
c
s,”
I
n
ter
n
a
ti
o
n
a
l
T
e
le
c
o
mm
u
n
ica
t
io
n
Un
io
n
-
T
e
lec
o
mm
u
n
ic
a
ti
o
n
st
a
n
d
a
rd
iza
ti
o
n
S
e
c
to
r
,
2
0
0
1
.
[1
1
]
J.
Be
n
e
st
y
,
S
.
M
a
k
in
o
,
J.
C
h
e
n
,
“
S
p
e
e
c
h
E
n
h
a
n
c
e
m
e
n
t,
”
S
p
rin
g
e
r
S
c
ien
c
e
&
Bu
sin
e
ss
M
e
d
ia,
M
a
r.
2
0
0
5
.
[1
2
]
S
.
Bo
ll
,
“
S
u
p
p
re
ss
io
n
o
f
A
c
o
u
stic
n
o
ise
i
n
S
p
e
e
c
h
u
sin
g
S
p
e
c
tral
S
u
b
t
ra
c
ti
o
n
,
”
IEE
E
T
r
a
n
sa
c
ti
o
n
s
o
n
Aco
u
stics
,
S
p
e
e
c
h
,
a
n
d
S
ig
n
a
l
P
ro
c
e
ss
in
g
,
v
o
l.
2
7
,
n
o
.
2
,
p
p
.
1
1
3
-
1
2
0
,
A
p
r.
1
9
7
9
.
[1
3
]
D.
W
a
n
g
,
J.
L
i
m
,
“
T
h
e
U
n
im
p
o
rtan
c
e
o
f
P
h
a
se
in
S
p
e
e
c
h
E
n
h
a
n
c
e
m
e
n
t,
”
IEE
E
T
ra
n
sa
c
ti
o
n
s o
n
Ac
o
u
stics
,
S
p
e
e
c
h
,
a
n
d
S
i
g
n
a
l
Pr
o
c
e
ss
in
g
, v
o
l.
3
0
,
n
o
.
4
,
A
u
g
.
1
9
8
2
.
[1
4
]
M
.
Be
ro
u
t
i,
R.
S
c
h
w
a
rtz,
J.
M
a
k
h
o
u
l,
“
En
h
a
n
c
e
m
e
n
t
o
f
S
p
e
e
c
h
Co
rr
u
p
ted
b
y
A
c
o
u
stic
n
o
ise
”
,
In
A
c
o
u
stics
,
S
p
e
e
c
h
,
a
n
d
S
ig
n
a
l
P
ro
c
e
ss
in
g
,
I
EE
E
In
ter
n
a
t
io
n
a
l
Co
n
f
e
re
n
c
e
o
n
ICA
S
S
P
'
7
9
,
v
o
l.
4
,
p
p
.
2
0
8
-
2
1
1
,
A
p
r.
1
9
7
9
.
[1
5
]
J.
S
.
L
im
,
A
.
V
.
Op
p
e
n
h
e
im
,
“
En
h
a
n
c
e
me
n
t
a
n
d
b
a
n
d
wid
t
h
C
o
mp
re
ss
io
n
o
f
n
o
isy
S
p
e
e
c
h
”
,
P
r
o
c
e
e
d
in
g
s
o
f
th
e
IEE
E,
V
o
l.
6
7
,
n
o
.
1
2
,
p
p
.
1
5
8
6
-
1
6
0
4
,
De
c
.
1
9
7
9
.
[1
6
]
T
.
S
.
G
u
n
a
w
a
n
,
“
A
u
d
io
c
o
m
p
re
ss
io
n
a
n
d
sp
e
e
c
h
e
n
h
a
n
c
e
m
e
n
t
u
sin
g
tem
p
o
ra
l
m
a
sk
in
g
m
o
d
e
ls”
,
Do
c
t
o
ra
l
d
isse
rtatio
n
,
T
h
e
Un
iv
e
rsity
o
f
Ne
w
S
o
u
th
W
a
les
,
2
0
0
7
.
[1
7
]
N.
V
irag
,
“
S
in
g
le
Ch
a
n
n
e
l
S
p
e
e
c
h
E
n
h
a
n
c
e
m
e
n
t
b
a
se
d
o
n
M
a
sk
in
g
P
ro
p
e
rti
e
s
o
f
th
e
Hu
m
a
n
A
u
d
it
o
ry
S
y
ste
m
,
”
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
S
p
e
e
c
h
a
n
d
Au
d
i
o
Pro
c
e
ss
in
g
,
v
o
l.
7
,
n
o
.
2
,
p
p
.
1
2
6
-
1
3
7
,
M
a
r.
1
9
9
9
.
[1
8
]
P
.
S
c
a
lart,
“
S
p
e
e
c
h
En
h
a
n
c
e
me
n
t
b
a
se
d
o
n
a
Pri
o
ri
S
i
g
n
a
l
t
o
n
o
i
se
E
stima
ti
o
n
”
,
ICA
S
S
P
-
9
6
,
IEE
E
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
,
1
9
9
6
.
[1
9
]
Y.
Ep
h
ra
im
,
D.
M
a
lah
,
“
S
p
e
e
c
h
En
h
a
n
c
e
m
e
n
t
u
sin
g
a
M
in
im
u
m
-
M
e
a
n
S
q
u
a
re
Err
o
r
S
h
o
rt
-
T
i
m
e
S
p
e
c
tral
Am
p
li
tu
d
e
E
stim
a
to
r
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Aco
u
stics
,
S
p
e
e
c
h
,
a
n
d
S
ig
n
a
l
Pro
c
e
ss
in
g
,
v
o
l.
3
2
,
n
o
.
6
,
p
p
.
1
1
0
9
-
1
1
2
1
,
De
c
.
1
9
8
4
.
[2
0
]
S.
Ra
n
g
a
c
h
a
ri,
P
.
C.
L
o
izo
u
,
“
A
No
ise
-
Esti
m
a
ti
o
n
A
lg
o
rit
h
m
f
o
r
Hig
h
ly
N
on
-
sta
ti
o
n
a
ry
En
v
iro
n
m
e
n
ts”
,
S
p
e
e
c
h
Co
mm
u
n
ica
ti
o
n
, v
o
l
.
4
8
,
n
o
.
2
,
p
p
.
2
2
0
-
2
3
1
,
F
e
b
.
2
0
0
6
.
[2
1
]
M.
A
.
F
ig
u
e
ired
o
,
R.
D.
No
w
a
k
,
S
.
J.
W
ri
g
h
t,
“
G
ra
d
ien
t
P
ro
jec
ti
o
n
f
o
r
S
p
a
rse
R
e
c
o
n
stru
c
ti
o
n
:
A
p
p
l
ica
ti
o
n
t
o
Co
m
p
re
ss
e
d
S
e
n
sin
g
a
n
d
o
th
e
r
In
v
e
rse
P
ro
b
lem
s”
,
IEE
E
J
o
u
rn
a
l
o
f
S
e
lec
ted
T
o
p
ics
in
S
i
g
n
a
l
Pro
c
e
ss
in
g
,
v
o
l.
1
,
n
o
.
4
,
n
o
.
5
8
6
-
5
9
7
,
De
c
.
2
0
0
7
.
[2
2
]
E.
Ro
t
h
a
u
se
r,
W
.
Ch
a
p
m
a
n
,
N.
G
u
tt
m
a
n
,
K.
No
rd
b
y
,
H.
S
il
b
ig
e
r,
G
.
Urb
a
n
e
k
,
M
.
W
e
in
sto
c
k
,
“
IEE
E
R
ec
o
m
m
e
n
d
e
d
P
ra
c
ti
c
e
f
o
r
S
p
e
e
c
h
Qu
a
li
ty
M
e
a
su
re
m
e
n
ts
”
,
IE
EE
T
ra
n
s.
A
u
d
i
o
El
e
c
tro
a
c
o
u
st
,
v
o
l.
1
7
,
n
o
.
3
,
p
p
.
2
2
5
-
2
4
6
,
S
e
p
.
1
9
9
6
.
[2
3
]
M
.
Be
ro
u
t
i,
R.
S
c
h
w
a
rt
z
,
J.
M
a
k
h
o
u
l,
“
En
h
a
n
c
e
me
n
t
o
f
S
p
e
e
c
h
Co
rr
u
p
ted
b
y
A
c
o
u
stic
n
o
ise
”
,
In
A
c
o
u
stics
,
S
p
e
e
c
h
,
a
n
d
S
ig
n
a
l
P
ro
c
e
ss
i
n
g
,
I
EE
E
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
ICA
S
S
P
'
7
9
,
Vo
l.
4
,
p
p
.
2
0
8
-
2
1
1
,
IEE
E
,
A
p
r.
1
9
7
9
.
[2
4
]
H.
G
u
sta
fss
o
n
,
S
.
E.
No
rd
h
o
lm
,
I.
Clae
ss
o
n
,
“
S
p
e
c
tral
S
u
b
trac
ti
o
n
u
sin
g
re
d
u
c
e
d
d
e
la
y
Co
n
v
o
lu
ti
o
n
a
n
d
A
d
a
p
ti
v
e
Av
e
ra
g
in
g
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s o
n
S
p
e
e
c
h
a
n
d
Au
d
io
Pro
c
e
s
sin
g
,
v
o
l.
9
,
n
o
.
8
,
p
p
.
7
9
9
-
8
0
7
,
No
v
.
2
0
0
1
.
[2
5
]
F
.
Ja
b
lo
u
n
,
B.
Ch
a
m
p
a
g
n
e
,
“
In
c
o
rp
o
ra
ti
n
g
th
e
Hu
m
a
n
He
a
rin
g
P
r
o
p
e
rti
e
s
i
n
th
e
S
ig
n
a
l
S
u
b
sp
a
c
e
A
p
p
ro
a
c
h
f
o
r
S
p
e
e
c
h
E
n
h
a
n
c
e
m
e
n
t
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
S
p
e
e
c
h
a
n
d
A
u
d
i
o
P
ro
c
e
ss
in
g
,
v
o
l.
1
1
,
n
o
.
6
,
p
p
.
7
0
0
-
7
0
8
,
N
o
v
.
2
0
0
3
.
[2
6
]
Y.
Hu
Y,
P
.
C
.
L
o
izo
u
,
“
S
p
e
e
c
h
En
h
a
n
c
e
m
e
n
t
b
a
se
d
o
n
W
a
v
e
let
T
h
re
sh
o
ld
in
g
th
e
M
u
lt
it
a
p
e
r
S
p
e
c
tru
m
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
S
p
e
e
c
h
a
n
d
Au
d
i
o
Pro
c
e
ss
in
g
,
v
o
l.
1
2
,
n
o
.
1
,
p
p
.
5
9
-
6
7
,
Ja
n
.
2
0
0
4
.
[2
7
]
Y.
Hu
Y,
P
.
C.
L
o
izo
u
,
“
In
c
o
rp
o
ra
ti
n
g
a
P
sy
c
h
o
a
c
o
u
stica
l
M
o
d
e
l
in
F
re
q
u
e
n
c
y
Do
m
a
in
S
p
e
e
c
h
En
h
a
n
c
e
m
e
n
t”,
IEE
E
S
i
g
n
a
l
Pro
c
e
ss
in
g
L
e
tt
e
rs
,
v
o
l.
1
1
,
n
o
.
2
,
p
p
.
2
7
0
-
2
7
3
,
F
e
b
2
0
0
4
.
[2
8
]
S
.
S
i
n
g
h
,
M
.
T
rip
a
t
h
y
,
R.
S
.
A
n
a
n
d
,
“
S
u
b
jec
ti
v
e
a
n
d
O
b
jec
ti
v
e
An
a
ly
sis
o
f
S
p
e
e
c
h
En
h
a
n
c
e
m
e
n
t
A
l
g
o
rit
h
m
s
f
o
r
S
in
g
le
Ch
a
n
n
e
l
S
p
e
e
c
h
P
a
tt
e
rn
s
o
f
In
d
ian
a
n
d
En
g
li
sh
L
a
n
g
u
a
g
e
s
”
,
IET
E
T
e
c
h
n
ica
l
Rev
iew
,
v
o
.
3
1
,
n
o
.
1
,
p
p
.
3
4
-
4
6
,
Ja
n
.
2
0
1
4
.
[2
9
]
P
.
Krish
n
a
m
o
o
rth
y
,
“
A
n
Ov
e
r
v
ie
w
o
f
S
u
b
jec
ti
v
e
a
n
d
Ob
jec
ti
v
e
Q
u
a
li
ty
M
e
a
su
re
s
f
o
r
No
isy
S
p
e
e
c
h
En
h
a
n
c
e
m
e
n
t
A
l
g
o
rit
h
m
s,” IE
T
E
Tec
h
n
ica
l
Re
v
ie
w
,
v
o
l.
2
8
,
n
o
.
4
,
p
p
.
2
9
2
-
3
0
1
,
Ju
l.
2
0
1
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.