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.
11
,
No
.
6
,
Dec
em
b
er
202
1
,
p
p
.
5
4
3
8
~
5
4
4
9
I
SS
N:
2088
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
11
i
6
.
pp
5
4
3
8
-
5
4
4
9
5438
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
E
m
o
tion r
ecog
nit
io
n f
ro
m
sy
lla
bic
u
nits
u
sing
k
-
nea
rest
-
neig
hbo
r
cla
ss
ific
a
tion a
nd
e
nerg
y
d
istribut
io
n
Abdella
h Ag
ri
m
a
1
,
I
l
ha
m
M
o
un
ir
2
,
Abdel
m
a
j
i
d F
a
rc
hi
3
,
L
a
ila
E
l
m
a
a
zo
uzi
4
,
B
a
dia
M
o
un
ir
5
1,
3
IM
M
I
L
a
b
o
ra
t
o
ry
,
F
a
c
u
lt
y
o
f
S
c
ien
c
e
s a
n
d
T
e
c
h
n
ics
,
Un
iv
e
rsity
Ha
ss
a
n
F
irst,
S
e
tt
a
t,
M
o
r
o
c
c
o
2,
4,
5
L
A
P
S
S
II
L
a
b
o
ra
to
ry
,
Hig
h
S
c
h
o
o
l
o
f
T
e
c
h
n
o
lo
g
y
,
Un
iv
e
rsit
y
C
a
d
i
Ay
y
a
d
,
S
a
f
i,
M
o
ro
c
c
o
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
l 2
1
,
2
0
20
R
ev
i
s
ed
A
p
r
2
6
,
2
0
2
1
A
cc
ep
ted
J
u
n
1
2
,
2
0
2
1
In
th
is
a
rti
c
le,
w
e
p
re
se
n
t
a
n
a
u
to
m
a
ti
c
te
c
h
n
iq
u
e
f
o
r
re
c
o
g
n
izin
g
e
m
o
ti
o
n
a
l
sta
tes
f
ro
m
sp
e
e
c
h
sig
n
a
ls.
T
h
e
m
a
in
f
o
c
u
s
o
f
th
is
p
a
p
e
r
is
to
p
re
se
n
t
a
n
e
ff
ici
e
n
t
a
n
d
re
d
u
c
e
d
se
t
o
f
a
c
o
u
stic
fe
a
tu
re
s
th
a
t
a
ll
o
ws
u
s
to
re
c
o
g
n
ize
th
e
f
o
u
r
b
a
sic
h
u
m
a
n
e
m
o
ti
o
n
s
(a
n
g
e
r,
sa
d
n
e
ss
,
jo
y
,
a
n
d
n
e
u
tral)
.
T
h
e
p
ro
p
o
se
d
f
e
a
tu
re
s
v
e
c
to
r
is
c
o
m
p
o
se
d
b
y
t
w
e
n
t
y
-
e
ig
h
t
m
e
a
su
re
m
e
n
ts
c
o
rre
sp
o
n
d
i
n
g
to
sta
n
d
a
rd
a
c
o
u
st
ic f
e
a
tu
re
s su
c
h
a
s
f
o
r
m
a
n
ts,
f
u
n
d
a
m
e
n
tal
f
re
q
u
e
n
c
y
(o
b
tai
n
e
d
b
y
P
ra
a
t
so
f
t
wa
re
)
a
s
w
e
ll
a
s
in
tro
d
u
c
in
g
n
e
w
f
e
a
tu
re
s
b
a
s
e
d
o
n
th
e
c
a
lcu
latio
n
o
f
th
e
e
n
e
rg
ies
in
so
m
e
sp
e
c
i
f
ic
f
re
q
u
e
n
c
y
b
a
n
d
s
a
n
d
t
h
e
ir
d
istri
b
u
ti
o
n
s
(th
a
n
k
s
to
M
ATLAB
c
o
d
e
s).
T
h
e
e
x
trac
ted
m
e
a
su
re
m
e
n
ts
a
re
o
b
tai
n
e
d
f
ro
m
s
y
ll
a
b
ic
u
n
it
s’
c
o
n
so
n
a
n
t/
v
o
w
e
l
(CV
)
d
e
riv
e
d
f
ro
m
M
o
ro
c
c
a
n
A
ra
b
ic
d
iale
c
t
e
m
o
ti
o
n
a
l
d
a
ta
b
a
se
(M
A
DED)
c
o
rp
u
s.
T
h
e
re
a
f
ter
,
th
e
d
a
ta
w
h
ich
h
a
s
b
e
e
n
c
o
ll
e
c
ted
is
th
e
n
train
e
d
b
y
a
k
-
n
ea
r
est
-
n
eig
h
b
o
r
(
KNN)
c
las
si
f
ier
to
p
e
rf
o
r
m
th
e
a
u
to
m
a
ted
re
c
o
g
n
it
io
n
p
h
a
se
.
T
h
e
re
su
lt
s
re
a
c
h
6
4
.
6
5
%
in
th
e
m
u
lt
i
-
c
las
s
c
la
ss
if
ic
a
ti
o
n
a
n
d
9
4
.
9
5
%
f
o
r
c
l
a
ss
if
i
c
a
ti
o
n
b
e
tw
e
e
n
p
o
siti
v
e
a
n
d
n
e
g
a
ti
v
e
e
m
o
ti
o
n
s.
K
ey
w
o
r
d
s
:
E
m
o
tio
n
r
ec
o
g
n
itio
n
KNN
Sp
ee
ch
s
i
g
n
a
l
S
y
llab
le
s
u
n
its
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
:
Ag
r
i
m
a
A
b
d
ellah
I
MI
I
L
ab
o
r
ato
r
y
Facu
lt
y
o
f
Sc
ien
ce
s
an
d
T
ec
h
n
ics
Un
i
v
er
s
it
y
Has
s
an
First
FS
T
o
f
Settat,
K
m
3
,
B
.
P
: 5
7
7
R
o
ad
o
f
C
asab
lan
ca
,
Set
tat,
Mo
r
o
cc
o
E
m
ail:
a
g
r
i
m
a.
ab
d
ella
h
@
g
m
ai
l.c
o
m
1.
I
NT
RO
D
UCT
I
O
N
I
n
ter
p
r
etin
g
e
m
o
tio
n
al
i
n
f
o
r
m
atio
n
i
s
i
m
p
er
ati
v
e
f
o
r
th
e
s
o
cial
in
ter
ac
tio
n
s
t
h
at
w
e
h
a
v
e
ev
er
y
d
a
y
[
1
]
,
[
2
]
.
T
h
is
in
v
o
lv
e
s
m
a
n
y
d
if
f
er
en
t
co
m
p
o
n
e
n
ts
s
u
c
h
as
b
o
d
y
la
n
g
u
a
g
e,
p
o
s
tu
r
e,
f
ac
ial
a
n
d
v
o
ca
l
ex
p
r
ess
io
n
s
.
An
y
i
n
f
o
r
m
atio
n
th
at
r
elate
s
t
o
th
e
s
e
co
m
p
o
n
e
n
ts
is
u
s
u
all
y
o
b
tain
ed
f
r
o
m
p
h
y
s
io
lo
g
ical
s
e
n
s
o
r
s
,
s
o
u
n
d
,
o
r
im
a
g
e
[
3
]
-
[
8
]
.
T
h
e
d
ata
is
m
a
n
ip
u
lated
o
f
a
v
e
r
y
lo
w
lev
e
l
(
s
o
u
n
d
s
a
m
p
les
o
r
ev
en
p
ix
els
o
f
i
m
a
g
es).
B
et
w
ee
n
t
h
i
s
lo
w
-
le
v
el
d
ata
an
d
t
h
e
i
n
ter
p
r
etatio
n
t
h
at
h
u
m
an
s
m
ak
e
o
f
it,
t
h
e
g
ap
is
e
n
o
r
m
o
u
s
.
I
n
d
ee
d
,
th
e
m
a
n
i
f
estat
io
n
o
f
e
m
o
tio
n
s
i
s
a
n
e
s
p
ec
iall
y
i
n
tr
icate
ar
ea
o
f
h
u
m
a
n
co
m
m
u
n
icatio
n
l
y
i
n
g
at
th
e
in
ter
s
ec
tio
n
o
f
m
u
lt
id
is
cip
li
n
a
r
y
s
cien
ce
s
s
u
ch
as
p
s
y
c
h
o
lo
g
y
,
p
s
y
ch
ia
tr
y
,
au
d
io
lo
g
y
,
an
d
co
m
p
u
ter
s
cie
n
ce
[
9
]
-
[
1
1
]
.
T
h
e
an
al
y
s
i
s
o
f
c
o
n
v
er
s
atio
n
s
‘
s
p
ee
c
h
an
a
l
y
t
i
cs’
,
i
s
o
n
e
o
f
t
h
e
r
ec
e
n
t
c
h
alle
n
g
e
s
i
n
m
a
n
y
ap
p
licatio
n
s
,
f
o
r
ex
a
m
p
le,
h
e
alth
m
o
n
ito
r
i
n
g
[
1
2
]
,
v
id
eo
g
a
m
es
[
1
3
]
,
an
d
co
m
p
u
ter
s
cie
n
ce
[
1
4
]
.
A
t
y
p
ica
l
d
o
m
ai
n
o
f
lear
n
in
g
e
m
o
tio
n
al
s
tate
f
r
o
m
an
a
l
y
s
is
o
f
c
o
n
v
er
s
atio
n
s
is
ca
ll
c
e
n
ter
s
.
I
n
d
ee
d
,
a
b
etter
u
n
d
er
s
ta
n
d
in
g
o
f
t
h
e
n
ee
d
s
o
f
a
cu
s
to
m
er
m
ea
n
s
f
o
r
th
e
en
t
er
p
r
is
e
a
b
etter
m
a
n
ag
e
m
e
n
t
an
d
g
r
ea
ter
b
en
ef
it
[
1
5
]
-
[
1
7
]
.
T
h
e
s
cien
ce
o
f
lear
n
in
g
la
y
s
o
n
d
ata
w
h
ic
h
co
m
e
u
s
u
all
y
f
r
o
m
d
if
f
er
en
t
s
ig
n
al
s
.
On
e
o
f
th
e
m
o
s
t
f
r
eq
u
en
tl
y
u
s
ed
s
i
g
n
als
is
t
h
e
s
p
ee
ch
s
ig
n
al.
Sp
ee
c
h
i
s
i
n
d
ee
d
o
n
e
o
f
t
h
e
f
u
n
d
a
m
en
tal
m
o
d
alities
t
h
at
m
e
n
u
s
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
E
mo
tio
n
r
ec
o
g
n
itio
n
fr
o
m
s
ylla
b
ic
u
n
its
u
s
in
g
…
(
A
b
d
ella
h
A
g
r
ima
)
5439
to
co
m
m
u
n
icate
.
Au
to
m
a
tic
s
p
ee
ch
r
ec
o
g
n
itio
n
s
y
s
te
m
s
g
i
v
e
th
e
m
ac
h
i
n
e
t
h
e
ab
ilit
y
to
tr
an
s
f
o
r
m
th
e
s
o
u
n
d
s
ig
n
al
in
to
a
s
er
ies
o
f
w
o
r
d
s
.
T
h
e
f
ield
o
f
au
to
m
atic
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
p
r
o
v
id
es
ac
ce
s
s
to
th
e
m
ea
n
i
n
g
o
f
th
is
s
er
ie
s
o
f
w
o
r
d
s
.
Star
tin
g
f
r
o
m
t
h
ese
to
o
ls
(
r
elativ
el
y
e
f
f
ec
tiv
e)
,
it
is
n
ec
ess
ar
y
to
g
o
f
u
r
th
er
.
T
h
e
q
u
esti
o
n
is
n
o
lo
n
g
er
m
er
el
y
ab
o
u
t
k
n
o
w
i
n
g
w
h
at
i
s
s
aid
,
b
u
t
also
to
k
n
o
w
t
h
e
co
n
te
x
t
o
f
t
h
e
p
r
o
n
u
n
c
iatio
n
o
f
th
e
s
en
te
n
ce
.
I
t
is
at
th
is
le
v
el
th
a
t
th
e
e
m
o
tio
n
a
l
d
i
m
en
s
io
n
i
n
t
er
v
en
es.
E
m
o
tio
n
r
ec
o
g
n
itio
n
is
a
d
if
f
ic
u
lt
er
r
an
d
s
in
ce
it
s
ex
p
r
ess
io
n
an
d
p
er
ce
p
tio
n
v
ar
y
e
x
tr
ao
r
d
in
ar
il
y
a
cr
o
s
s
cu
lt
u
r
es,
s
p
o
k
e
n
lan
g
u
a
g
es,
an
d
s
e
n
te
n
ce
s
[
1
8
]
-
[
2
1
]
.
L
ik
e
an
y
e
m
e
r
g
i
n
g
f
ield
,
it
h
a
s
n
't
b
ee
n
co
m
p
lete
l
y
w
ell
-
es
tab
lis
h
ed
y
et,
b
u
t
to
q
u
alif
y
e
m
o
tio
n
al
s
p
ee
ch
,
r
esear
c
h
er
s
h
av
e
atte
m
p
ted
to
r
ep
r
esen
t
it
b
y
v
ar
io
u
s
d
e
s
cr
ip
to
r
s
r
elate
d
to
s
p
o
k
en
co
n
ten
t
(
I
f
w
e
d
o
n
o
t
tak
e
in
to
ac
co
u
n
t
th
e
v
ar
iatio
n
o
f
th
e
s
e
n
ten
ce
,
i
t
is
d
if
f
icu
lt
to
d
if
f
er
e
n
tiate
b
et
w
e
en
a
li
n
g
u
i
s
tic
a
n
d
e
m
o
tio
n
al
v
ar
iatio
n
)
,
ar
ticu
lat
o
r
y
p
o
s
it
io
n
(
p
o
s
itio
n
o
f
t
h
e
j
a
w
ac
co
r
d
in
g
to
th
e
p
r
o
n
o
u
n
ce
d
e
m
o
tio
n
)
[
2
2
]
,
an
d
ac
o
u
s
tic
p
r
o
p
er
ties
[
2
3
]
,
[
2
4
]
.
T
h
e
co
n
tr
ib
u
tio
n
w
e
ar
e
p
r
ese
n
ti
n
g
h
er
e
w
i
s
h
to
g
ai
n
an
u
n
d
er
s
tan
d
in
g
o
f
t
h
e
s
u
b
j
ec
t
an
d
ex
p
o
s
in
g
an
ex
p
lo
r
ato
r
y
asp
ec
t
o
f
it.
Fir
s
t,
u
n
li
k
e
th
e
m
aj
o
r
ity
o
f
t
h
e
e
x
is
t
in
g
w
o
r
k
,
w
e
w
ill
co
n
s
id
er
A
r
ab
ic
p
h
o
n
e
m
es
C
V
(
p
lo
s
i
v
e
co
n
s
o
n
a
n
t/
v
o
w
e
l
)
in
s
tead
o
f
w
h
o
le
s
e
n
te
n
ce
s
.
Seco
n
d
,
w
e
w
il
l
lo
o
k
at
d
ata
ac
o
u
s
tic
f
ea
t
u
r
es
i
n
t
er
m
s
o
f
e
n
er
g
y
a
n
d
it
s
d
is
tr
ib
u
tio
n
i
n
s
o
m
e
s
p
ec
i
f
ic
b
a
n
d
s
.
Fin
a
ll
y
,
to
lear
n
f
r
o
m
t
h
e
co
n
s
tr
u
cted
d
ata
s
e
ts
an
d
p
r
ed
ict
e
m
o
tio
n
al
s
tate
o
f
s
p
ea
k
er
,
w
e
u
s
e
t
h
e
i
n
s
ta
n
ce
-
b
ased
class
i
f
icat
io
n
al
g
o
r
ith
m
s
k
-
n
ea
r
est
-
n
ei
g
h
b
o
r
(
KNN)
.
Ov
er
th
e
p
ast
f
i
f
tee
n
y
ea
r
s
,
i
n
cr
ea
s
i
n
g
n
u
m
b
er
o
f
r
esear
c
h
er
s
h
av
e
b
ee
n
in
ter
ested
in
t
h
e
s
tu
d
y
o
f
e
m
o
tio
n
s
in
s
p
ee
ch
[
2
5
]
.
Fro
m
d
ata
lab
elled
e
m
o
tio
n
all
y
,
a
f
ea
tu
r
e
s
et
b
ased
o
n
s
p
ec
tr
o
-
te
m
p
o
r
al
f
ea
t
u
r
es (
f
o
r
in
s
ta
n
ce
p
itc
h
,
i
n
ten
s
it
y
,
an
d
en
er
g
y
o
f
th
e
s
p
ee
ch
t
h
at
a
r
e
ex
tr
ac
ted
u
s
in
g
al
g
o
r
ith
m
s
)
is
d
ev
elo
p
ed
f
o
r
p
r
o
ce
s
s
in
g
t
h
e
v
o
ice
s
i
g
n
al.
Su
p
r
a
-
s
eg
m
e
n
tal
r
ep
r
esen
tat
io
n
s
(
also
ca
lled
lo
w
-
lev
el
d
es
cr
ip
to
r
s
(
L
L
D)
)
ar
e
d
er
iv
ed
f
r
o
m
li
n
g
u
i
s
tic
u
n
it
s
s
u
ch
as
s
e
n
te
n
ce
s
,
w
o
r
d
s
,
o
r
p
h
o
n
e
m
e
s
[
2
6
]
,
[
2
7
]
.
Hig
h
-
le
v
el
d
escr
ip
to
r
s
(
HL
D)
s
u
c
h
a
s
i
n
[
2
8
]
-
[
3
0
]
ar
e
co
m
m
o
n
l
y
ex
tr
ac
ted
b
y
co
m
p
u
ti
n
g
v
ar
io
u
s
s
ta
tis
tic
s
f
r
o
m
t
h
e
L
L
Ds
o
v
er
th
e
d
ef
in
ed
lin
g
u
i
s
tic
u
n
it
s
.
T
h
e
s
ize
o
f
t
h
e
f
ea
t
u
r
e
s
et
e
x
ce
ed
s
t
h
o
u
s
a
n
d
s
(
H
L
D
a
n
d
L
L
D)
co
n
ti
n
g
e
n
t
u
p
o
n
t
h
e
n
u
m
b
er
o
f
s
tatis
t
ics
ex
tr
ac
ted
.
T
h
e
d
is
cr
i
m
i
n
ati
n
g
cla
s
s
i
f
ie
r
s
ar
e
th
e
n
tr
ain
ed
o
n
th
ese
h
i
g
h
-
lev
el
ch
ar
ac
ter
is
tic
s
f
o
r
a
m
u
ltit
u
d
e
o
f
ta
s
k
s
s
u
ch
a
s
b
in
ar
y
(
v
ale
n
ce
,
ac
ti
v
atio
n
)
o
r
ca
t
eg
o
r
ical
(
h
ap
p
y
,
s
ad
,
j
o
y
,
s
ad
,
etc.
)
class
i
f
icatio
n
s
as
w
ell
as
r
e
g
r
ess
io
n
o
n
co
n
ti
n
u
o
u
s
e
m
o
tio
n
al
attr
ib
u
tes.
M
an
y
clas
s
i
f
icatio
n
ap
p
r
o
ac
h
es
h
av
e
b
ee
n
ap
p
lied
to
em
o
tio
n
r
ec
o
g
n
itio
n
s
y
s
te
m
s
(
SER).
I
n
d
ee
d
,
th
e
h
y
b
r
id
Gau
s
s
ia
n
m
ix
t
u
r
e
m
o
d
el
(
GM
M)
,
n
eu
r
al
n
et
w
o
r
k
w
it
h
(
R
B
F,
P
NN,
SVM,
E
L
M)
[
3
1
]
,
[
3
2
]
,
an
d
h
id
d
en
Ma
r
k
o
v
m
o
d
el
(
HM
M)
[
3
3
]
w
er
e
u
s
ed
to
g
e
n
er
ate
a
m
o
d
el
o
f
r
ec
o
g
n
itio
n
b
y
e
m
o
t
i
o
n
an
d
b
y
g
e
n
d
er
(
m
a
n
/
w
o
m
a
n
)
o
f
th
e
s
p
ea
k
er
.
A
l
s
o
,
a
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
i
n
e
(
SVM)
co
m
b
i
n
ed
w
it
h
a
p
o
l
y
n
o
m
ial
n
u
c
leu
s
w
as
ap
p
lie
d
f
o
r
m
u
lt
i
-
cla
s
s
a
n
d
m
u
lti
-
co
r
p
u
s
r
ec
o
g
n
itio
n
s
y
s
te
m
s
[
3
4
]
,
[
3
5
]
.
R
ec
en
tl
y
,
d
ee
p
n
eu
r
al
n
et
w
o
r
k
s
(
DN
Ns)
[
3
6
]
h
av
e
b
ee
n
d
esi
g
n
e
d
to
ad
d
r
ess
SER
p
r
o
b
lem
s
.
O
n
e
o
f
t
h
e
ad
v
a
n
ta
g
es
o
f
th
i
s
ap
p
r
o
ac
h
is
tr
an
s
f
e
r
lear
n
in
g
.
T
r
an
s
f
er
lear
n
in
g
c
o
n
s
is
ts
o
f
lear
n
i
n
g
th
e
f
ir
s
t la
y
er
s
o
f
th
e
n
et
w
o
r
k
p
er
f
o
r
m
in
g
j
o
in
t o
p
er
atio
n
s
f
r
o
m
o
n
e
co
r
p
u
s
to
an
o
t
h
er
,
o
r
f
r
o
m
o
n
e
e
m
o
tio
n
to
an
o
th
er
-
o
n
m
o
r
e
d
at
a
th
a
n
t
h
e
d
ee
p
er
lay
er
s
a
s
s
i
g
n
ed
to
a
s
p
ec
if
ic
ta
s
k
[
3
7
]
.
Kay
a
a
n
d
Kar
p
o
v
[
3
8
]
u
s
ed
k
er
n
el
e
x
tr
e
m
e
lear
n
in
g
m
ac
h
in
es
(
E
L
M)
,
ad
ap
ted
to
a
b
ase
o
f
f
e
w
s
a
m
p
les
f
o
r
m
an
y
c
h
ar
ac
ter
is
tic
s
.
T
h
e
y
also
p
r
o
p
o
s
ed
a
n
e
w
m
eth
o
d
o
f
n
o
r
m
alizi
n
g
t
h
e
c
h
ar
ac
ter
is
ti
cs
as
a
n
a
lter
n
ati
v
e
to
th
e
s
ta
n
d
ar
d
n
o
r
m
al
izatio
n
o
f
ce
n
ter
i
n
g
-
r
ed
u
ctio
n
.
T
o
s
u
m
m
ar
ize
th
e
s
tate
-
of
-
th
e
-
ar
t o
f
th
e
f
ie
ld
,
w
e
r
ef
er
t
h
e
r
ea
d
er
to
T
ab
le
1
.
I
n
o
u
r
s
tu
d
y
,
w
e
i
n
v
e
s
ti
g
ate
t
h
e
u
til
izatio
n
o
f
s
p
ee
c
h
s
i
g
n
al
s
f
o
r
d
etec
tin
g
e
m
o
tio
n
.
W
e
p
r
o
p
o
s
e
an
e
m
o
tio
n
r
ec
o
g
n
itio
n
m
eth
o
d
f
r
o
m
v
o
ca
l
d
ata
b
ased
o
n
a
s
i
g
n
al
p
r
o
ce
s
s
i
n
g
al
g
o
r
ith
m
as
s
h
o
w
n
i
n
Fi
g
u
r
e
1
.
Ou
r
m
et
h
o
d
co
n
s
is
t
s
o
f
th
r
ee
s
tep
s
:
(
1
)
d
ata
p
r
ep
a
r
atio
n
(
2
)
s
p
ee
ch
f
ea
t
u
r
es
ex
tr
ac
tio
n
a
s
s
h
o
w
n
i
n
T
ab
le
2
,
an
d
(
3
)
class
if
icat
io
n
u
tili
zi
n
g
th
e
KNN
al
g
o
r
ith
m
.
Fig
u
r
e
1
.
Glo
b
al
s
p
ee
ch
e
m
o
ti
o
n
r
ec
o
g
n
itio
n
s
y
s
te
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
6
,
Dec
em
b
er
2021
:
5
4
3
8
-
5
4
4
9
5440
T
ab
le
1
.
A
b
r
ief
s
u
m
m
ar
y
o
f
S
E
R
w
ith
d
i
f
f
er
en
t d
atab
ases
,
d
if
f
er
en
t
f
ea
t
u
r
es a
n
d
clas
s
i
f
ica
tio
n
alg
o
r
it
h
m
s
R
e
f
e
r
e
n
c
e
s
D
a
t
a
se
t
F
e
a
t
u
r
e
s
M
o
d
e
l
s/
C
l
a
ssi
f
i
e
r
s
B
e
st
r
e
su
l
t
H
a
n
e
t
a
l
.
[
3
9
]
I
EM
O
C
A
P
-
DB
P
i
t
c
h
-
b
a
se
d
f
e
a
t
u
r
e
s,
a
n
d
t
h
e
i
r
d
e
l
t
a
f
e
a
t
u
r
e
a
c
r
o
ss
t
i
me
f
r
a
me
s.
M
F
C
C
C
o
e
f
f
i
c
i
e
n
t
s,
D
N
N
a
n
d
Ex
t
r
e
me
L
e
a
r
n
i
n
g
M
a
c
h
i
n
e
5
4
.
3
0
%
a
v
e
r
a
g
e
r
e
c
o
g
n
i
t
i
o
n
r
a
t
e
.
D
e
n
g
e
t
a
l
.
[
4
0
]
A
B
C
D
B
,
A
I
B
O
D
B
,
S
U
S
A
S
DB
L
o
w
-
L
e
v
e
l
D
e
scri
p
t
o
r
s
D
e
n
o
i
si
n
g
a
u
t
o
-
e
n
c
o
d
e
r
s
a
n
d
S
V
M
6
4
.
1
8
%
r
e
c
o
g
n
i
t
i
o
n
r
a
t
e
f
o
r
A
B
C
D
B
.
6
2
.
7
4
%
a
v
e
r
a
g
e
r
e
c
o
g
n
i
t
i
o
n
r
a
t
e
f
o
r
S
U
S
A
S
D
B
.
M
i
r
sa
ma
d
i
e
t
a
l
.
[
4
1
]
I
EM
O
C
A
P
c
o
r
p
u
s
A
u
t
o
mat
i
c
a
l
l
y
l
e
a
r
n
e
d
b
y
D
R
N
N
,
a
s
w
e
l
l
a
s h
a
n
d
-
c
r
a
f
t
e
d
L
L
D
s
c
o
n
si
st
i
n
g
o
f
F
0
,
v
o
i
c
i
n
g
p
r
o
b
a
b
i
l
i
t
y
,
f
r
a
me
e
n
e
r
g
y
,
Z
C
R
,
a
n
d
M
F
C
C
D
e
e
p
R
N
N
P
r
o
p
o
se
d
sy
st
e
m
w
i
t
h
r
a
w
sp
e
c
t
r
a
l
f
e
a
t
u
r
e
s h
a
s
6
1
.
8
%
r
e
c
o
g
n
i
t
i
o
n
r
a
t
e
.
P
r
o
p
o
se
d
sy
st
e
m
w
i
t
h
L
L
D
f
e
a
t
u
r
e
s h
a
s
6
3
.
5
%
r
e
c
o
g
n
i
t
i
o
n
r
a
t
e
.
M
a
o
e
t
a
l
.
[
4
2
]
S
A
V
EE
D
B
,
B
e
r
l
i
n
EM
O
D
B
,
D
ES D
B
,
M
ES
D
B
A
u
t
o
mat
i
c
a
l
l
y
l
e
a
r
n
e
d
b
y
C
N
N
C
N
N
7
3
.
6
%
a
c
c
u
r
a
c
y
f
o
r
S
A
V
EE
D
B
.
7
9
.
9
%
f
o
r
D
ES D
B
.
7
8
.
3
%
f
o
r
M
ES D
B
.
8
5
.
2
%
f
o
r
E
M
O
D
B
.
I
ssa,
D
i
a
s,
e
t
a
l
.
[
4
3
]
R
A
V
D
ESS
,
B
e
r
l
i
n
EM
O
D
B
,
I
EM
O
C
A
P
M
e
l
-
f
r
e
q
u
e
n
c
y
c
e
p
st
r
a
l
c
o
e
f
f
i
c
i
e
n
t
s (
M
F
C
C
s)
,
C
h
r
o
ma
-
g
r
a
m,
M
e
l
-
sca
l
e
d
sp
e
c
t
r
o
g
r
a
m,
S
p
e
c
t
r
a
l
c
o
n
t
r
a
st
f
e
a
t
u
r
e
,
T
o
n
n
e
t
z
r
e
p
r
e
se
n
t
a
t
i
o
n
.
D
e
e
p
C
N
N
7
1
.
6
1
%
f
o
r
R
A
V
D
ESS
D
B
w
i
t
h
8
c
l
a
sse
s.
8
6
.
1
%
f
o
r
E
M
O
-
D
B
7
c
l
a
sse
s.
9
5
.
7
1
%
f
o
r
EM
O
-
D
B
w
i
t
h
7
c
l
a
sse
s.
6
4
.
3
%
f
o
r
I
EM
O
C
A
P
w
i
t
h
4
c
l
a
sse
s.
S
i
n
i
t
h
e
t
a
l
.
[
4
4
]
B
e
r
l
i
n
-
Emo
S
A
V
EE
P
i
t
c
h
,
i
n
t
e
n
s
i
t
y
,
M
F
C
C
S
u
p
p
o
r
t
V
e
c
t
o
r
R
e
g
r
e
ssi
o
n
M
a
l
e
s:
6
7
.
5
%.
F
e
mal
e
s
:
7
0
%
.
B
o
t
h
:
7
5
%,
6
1
.
2
5
%
2.
P
RO
P
O
SE
D
M
E
T
H
O
D
I
n
th
i
s
ar
ticle,
a
n
e
w
c
h
ar
ac
t
er
is
tic
ex
tr
ac
tio
n
s
ch
e
m
e
is
p
r
o
p
o
s
ed
w
h
ic
h
is
b
ased
o
n
a
p
s
eu
d
o
-
p
h
o
n
etic
ap
p
r
o
ac
h
[
4
5
]
,
[
4
6
]
.
T
h
e
k
e
y
p
o
in
t
o
f
o
u
r
w
o
r
k
i
s
t
o
ex
tr
ac
t
th
e
ch
ar
ac
ter
i
s
tics
ac
co
r
d
in
g
to
d
if
f
er
e
n
t
s
eg
m
e
n
ts
s
u
ch
as
t
h
e
s
y
l
lab
ic
u
n
it
s
to
r
em
ed
y
t
h
e
lin
g
u
is
t
ic
v
ar
iatio
n
co
n
s
tr
ain
t.
T
h
ese
s
e
g
m
en
t
s
ar
e
id
en
ti
f
ied
b
y
m
a
n
u
al
s
eg
m
e
n
t
atio
n
o
f
t
h
e
s
p
ee
ch
s
i
g
n
al.
O
u
r
d
ev
elo
p
ed
m
et
h
o
d
is
b
ased
o
n
ex
tr
ac
tin
g
c
lu
e
s
u
tili
zi
n
g
s
i
g
n
a
l
p
r
o
ce
s
s
in
g
m
eth
o
d
s
.
L
o
w
-
lev
e
l
d
escr
ip
to
r
s
(
L
L
D
s
)
an
d
h
i
g
h
-
le
v
el
d
esc
r
ip
to
r
s
(
HL
Ds)
as
s
h
o
w
n
i
n
T
ab
le
2
ar
e
o
b
tain
ed
f
r
o
m
a
v
o
ice
s
ig
n
al
lab
el
led
b
y
f
o
u
r
e
m
o
tio
n
s
(
a
n
g
er
,
s
ad
n
es
s
,
j
o
y
,
a
n
d
n
eu
tr
al)
.
E
ac
h
c
h
o
s
e
n
a
u
d
io
s
e
q
u
en
ce
m
u
s
t
f
ir
s
tl
y
s
atis
f
y
t
h
e
au
d
ib
ilit
y
cr
iter
io
n
an
d
t
h
er
ea
f
ter
p
ass
es
th
r
o
u
g
h
th
e
f
o
llo
w
i
n
g
p
r
o
ce
s
s
:
Mo
d
u
latin
g
th
e
s
ig
n
al
ac
co
r
d
in
g
to
a
s
et
o
f
co
n
tex
tu
al,
c
u
lt
u
r
al,
an
d
lin
g
u
i
s
tic
v
ar
iab
le
s
w
h
o
s
e
p
u
r
p
o
s
e
is
to
allo
w
co
m
m
u
n
ica
tio
n
(
e
m
o
tio
n
al
o
r
n
o
t)
,
C
ap
tu
r
i
n
g
t
h
e
s
i
g
n
al
p
r
o
d
u
ce
d
b
y
t
h
e
s
p
ea
k
er
,
An
n
o
tatin
g
t
h
e
s
i
g
n
a
l [
4
7
]
,
Seg
m
en
tin
g
t
h
e
s
i
g
n
al
in
s
y
l
la
b
ic
f
o
r
m
at,
E
x
tr
ac
tin
g
ac
o
u
s
tic
f
ea
tu
r
e
s
b
y
u
s
in
g
s
ig
n
al
p
r
o
ce
s
s
i
n
g
tec
h
n
iq
u
e
s
,
C
las
s
i
f
y
i
n
g
u
s
in
g
K
-
NN
al
g
o
r
ith
m
.
T
ab
le
2
.
Hig
h
an
d
lo
w
-
le
v
el
s
ettin
g
s
f
o
r
th
e
e
m
o
t
io
n
r
ec
o
g
n
itio
n
s
y
s
te
m
L
o
w
-
l
e
v
e
l
d
e
scri
p
t
o
r
s
H
i
g
h
-
l
e
v
e
l
d
e
scri
p
t
o
r
s
T
h
e
f
o
u
r
f
i
r
st
f
o
r
man
t
s
I
n
t
e
n
si
t
y
P
i
t
c
h
Ji
t
t
e
r
(
L
o
c
a
l
,
A
b
so
l
u
t
e
)
S
h
i
mm
e
r
(
L
o
c
a
l
,
a
b
so
l
u
t
e
)
P
u
l
se
s
T
h
e
e
n
e
r
g
y
i
n
t
h
e
si
x
b
a
n
d
s (
1
,
2
,
3
,
3
,
4
,
5
,
6
)
P
e
r
c
e
n
t
a
g
e
o
f
e
n
e
r
g
y
i
n
t
h
e
s
i
x
b
a
n
d
s
(
1
,
2
,
3
,
3
,
4
,
5
,
6
)
S
t
a
n
d
a
r
d
d
e
v
i
a
t
i
o
n
M
e
d
i
a
n
p
i
t
c
h
M
e
a
n
p
i
t
c
h
M
a
x
i
m
u
m
p
i
t
c
h
M
i
n
i
m
u
m
p
i
t
c
h
2.
1
.
Da
t
a
prepa
ra
t
i
o
n
An
y
s
cie
n
ti
f
ic
s
t
u
d
y
i
n
m
ac
h
i
n
e
lear
n
i
n
g
is
ex
tr
e
m
el
y
d
ep
en
d
en
t
o
n
t
h
e
d
ata
th
at
i
s
u
s
ed
to
d
escr
ib
e
th
e
p
h
e
n
o
m
e
n
o
n
to
b
e
m
o
d
elled
.
T
h
er
ef
o
r
e,
th
e
co
llectio
n
o
f
in
f
o
r
m
at
io
n
ad
ap
ted
to
th
e
task
t
h
at
w
e
w
an
t
to
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
E
mo
tio
n
r
ec
o
g
n
itio
n
fr
o
m
s
ylla
b
ic
u
n
its
u
s
in
g
…
(
A
b
d
ella
h
A
g
r
ima
)
5441
m
o
d
el
b
ec
o
m
e
s
a
m
aj
o
r
is
s
u
e
f
o
r
o
b
tain
in
g
a
d
et
ec
tio
n
m
o
d
el
w
it
h
a
s
u
f
f
ic
ien
t
l
y
s
tr
o
n
g
g
e
n
er
aliza
tio
n
p
o
w
er
.
No
w
ad
a
y
s
t
h
er
e
h
a
s
b
ee
n
s
o
m
e
g
en
u
i
n
e
w
o
r
k
in
th
e
zo
n
e
o
f
e
m
o
t
io
n
r
ec
o
g
n
itio
n
i
n
g
en
er
al
an
d
e
m
o
tio
n
f
r
o
m
s
o
u
n
d
s
p
ec
i
f
icall
y
;
h
o
w
e
v
er
,
an
en
o
r
m
o
u
s
p
o
r
tio
n
o
f
t
h
is
w
o
r
k
h
as
b
ee
n
as
s
es
s
ed
o
n
ac
ted
s
p
ee
ch
[
4
8
]
,
[
4
9
]
,
an
d
v
er
y
litt
le
w
o
r
k
h
as
b
ee
n
d
o
n
e
o
n
r
ea
l a
n
d
s
p
o
n
tan
eo
u
s
s
p
ee
ch
[
5
0
]
.
T
h
e
s
tu
d
y
p
r
ese
n
ted
i
n
t
h
is
ar
ticle
is
lo
ca
ted
i
n
t
h
e
co
n
te
x
t
o
f
e
m
o
tio
n
d
etec
tio
n
d
u
r
in
g
i
n
ter
ac
tio
n
s
b
et
w
ee
n
Mo
r
o
cc
an
citize
n
s
,
a
g
ed
1
6
–
6
0
y
ea
r
s
ex
p
r
es
s
i
n
g
f
o
u
r
b
a
s
ic
e
m
o
t
io
n
s
: h
ap
p
in
e
s
s
,
an
g
er
,
n
eu
tr
al
s
tate,
an
d
s
ad
n
ess
.
O
u
r
co
r
p
u
s
Mo
r
o
cc
an
A
r
ab
ic
d
ialec
t
e
m
o
tio
n
al
d
atab
ase
(
M
A
DE
D
)
is
o
b
tai
n
ed
f
r
o
m
u
n
co
n
tr
o
lled
r
ec
o
r
d
in
g
s
,
co
ll
ec
ted
f
r
o
m
r
ea
l
s
it
u
atio
n
s
t
h
a
t
ca
n
b
e
ex
tr
e
m
el
y
d
i
v
er
s
e.
T
h
e
s
elec
ted
s
u
b
s
et
in
cl
u
d
es
s
it
u
atio
n
s
tak
i
n
g
p
lace
in
d
if
f
er
e
n
t
co
n
tex
ts
(
in
d
o
o
r
,
o
u
td
o
o
r
s
ce
n
es,
m
o
n
o
lo
g
u
e,
an
d
d
ialo
g
u
e)
.
T
h
e
e
m
o
tio
n
s
ar
e
v
alid
ated
a
n
d
la
b
elled
b
y
t
h
e
i
n
ter
f
ac
e
s
h
o
w
n
in
Fig
u
r
e
2
b
u
ilt
b
y
o
u
r
tea
m
.
Firstl
y
,
t
h
e
d
ata
s
et
is
ch
a
n
g
ed
o
v
er
to
.
w
av
f
o
r
m
a
t
an
d
cu
t
in
to
a
s
y
llab
ic
s
tr
u
ct
u
r
e
as
b
ein
g
t
h
e
b
asic
u
n
it
s
i
m
ilar
to
th
e
ess
en
tial
u
n
i
t
o
f
o
u
r
h
a
n
d
li
n
g
.
T
h
en
it
g
o
es
t
h
r
o
u
g
h
a
n
o
th
er
cla
s
s
i
f
ic
atio
n
s
tep
o
f
t
h
e
s
y
llab
ic
t
y
p
e
s
to
r
ed
in
t
h
e
s
a
m
e
f
o
ld
er
.
Fig
u
r
e
2
.
Desk
to
p
an
n
o
tatio
n
t
o
o
l u
s
ed
f
o
r
e
m
o
tio
n
e
v
alu
a
tio
n
I
n
th
e
liter
atu
r
e,
th
er
e
is
a
g
r
e
at
d
ea
l
o
f
d
is
cu
s
s
io
n
o
n
t
h
e
len
g
t
h
o
f
t
h
e
au
d
io
f
o
r
w
h
ich
t
h
e
em
o
tio
n
ca
n
b
e
ex
tr
ac
ted
d
ep
en
d
ab
ly
[
5
1
]
-
[
5
4
]
.
I
n
o
u
r
s
tu
d
y
,
w
e
p
r
o
p
o
s
e
a
b
asic
m
et
h
o
d
o
lo
g
y
f
o
r
s
eg
m
e
n
ti
n
g
t
h
e
au
d
io
b
ased
o
n
a
p
s
eu
d
o
-
p
h
o
n
etic
ap
p
r
o
ac
h
.
T
h
e
m
ain
id
e
a
is
to
ex
tr
ac
t
s
o
m
e
f
ea
t
u
r
es
l
ik
e
(
f
o
r
m
a
n
ts
,
p
itc
h
,
an
d
en
er
g
ies
i
n
s
ix
b
an
d
s
)
f
r
o
m
t
h
e
s
p
ee
c
h
s
i
g
n
al
ac
co
r
d
in
g
to
th
e
s
y
llab
ic
s
e
g
m
e
n
t
(
C
V)
,
w
h
er
e
C
in
d
icate
s
C
o
n
s
o
n
an
t a
n
d
V
Vo
w
el.
Fo
r
all
o
u
r
ex
p
er
i
m
e
n
ts
,
w
e
co
n
s
id
er
ed
o
n
ly
f
o
u
r
e
m
o
tio
n
s
,
in
p
a
r
ticu
lar
,
j
o
y
,
an
g
er
,
s
ad
n
es
s
,
an
d
n
eu
tr
al
b
ec
au
s
e
t
h
e
r
u
le
-
b
ased
e
m
o
tio
n
ex
tr
ac
ti
o
n
s
y
s
te
m
t
h
at
w
e
u
s
ed
ca
ter
ed
to
o
n
l
y
th
e
s
e
f
o
u
r
e
m
o
tio
n
s
.
T
h
er
e
w
er
e
9
7
9
s
o
u
n
d
r
ec
o
r
d
s
r
elatin
g
to
th
e
s
e
4
e
m
o
tio
n
s
,
s
p
ec
if
ica
ll
y
2
5
0
s
y
l
lab
les
f
o
r
/b
a/,
2
4
0
f
o
r
/Du
/,
2
7
0
f
o
r
/Ki/,
an
d
2
1
9
f
o
r
/T
a/
as sh
o
w
n
in
T
ab
le
3
.
T
ab
le
3
.
Nu
m
b
er
o
f
au
d
io
f
ile
s
co
r
r
esp
o
n
d
in
g
to
4
e
m
o
tio
n
s
V
a
r
i
a
b
l
e
N
J
S
A
/b
a/
51
63
66
70
/Du
/
62
62
50
66
/Ki/
70
60
54
86
/T
a/
54
50
54
61
N
o
t
e
:
N
:
n
e
u
t
r
a
l
,
J:
j
o
y
,
S
:
sad
n
e
ss,
A
:
a
n
g
e
r
2
.
2
.
Sp
ee
ch
f
ea
t
ures e
x
t
ra
ct
i
o
n
A
t
p
r
esen
t,
th
er
e
is
n
o
a
g
r
ee
m
en
t
o
n
t
h
e
b
est
ar
r
an
g
e
m
e
n
t
o
f
i
m
p
o
r
tan
t
d
e
s
cr
ip
to
r
s
f
o
r
a
n
au
to
m
at
ic
e
m
o
tio
n
r
ec
o
g
n
i
tio
n
s
y
s
te
m
.
T
h
e
m
o
s
t
w
ell
-
k
n
o
w
n
p
r
ac
ti
ce
is
to
s
elec
t
a
lar
g
e
n
u
m
b
er
to
h
a
v
e
a
r
ic
h
er
class
i
f
icatio
n
.
Ho
w
e
v
er
,
in
cr
ea
s
in
g
t
h
e
n
u
m
b
er
o
f
d
escr
i
p
to
r
s
to
o
h
ig
h
f
o
r
a
co
r
p
u
s
o
f
r
ed
u
ce
d
s
ize
ca
n
p
o
s
s
ib
l
y
lead
to
p
e
r
f
o
r
m
a
n
ce
d
eg
r
ad
atio
n
an
d
ac
co
r
d
in
g
l
y
b
e
co
u
n
ter
p
r
o
d
u
ctiv
e.
T
o
s
o
lv
e
th
is
p
r
o
b
lem
,
it
is
n
ec
es
s
ar
y
to
ad
o
p
t
a
n
e
w
s
tr
ateg
y
th
a
t
w
ill
m
a
k
e
it
p
o
s
s
i
b
le
to
r
ed
u
ce
th
e
n
u
m
b
er
o
f
d
escr
ip
to
r
s
w
h
ile
k
ee
p
in
g
an
ac
ce
p
tab
le
r
ec
o
g
n
i
tio
n
r
ate.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
6
,
Dec
em
b
er
2021
:
5
4
3
8
-
5
4
4
9
5442
T
h
e
s
y
s
te
m
w
e
p
r
o
p
o
s
e
is
co
m
p
o
s
ed
o
f
s
ta
n
d
ar
d
ac
o
u
s
tic
f
ea
tu
r
es
a
s
s
h
o
w
n
i
n
T
ab
le
2
th
at
s
er
v
ed
a
s
th
e
ch
alle
n
g
e
b
aselin
e
s
e
t
s
in
ce
.
T
h
e
in
ter
s
p
ee
ch
e
m
o
tio
n
ch
alle
n
g
e
2
0
0
9
[
5
5
]
.
T
h
e
n
o
v
e
lt
y
co
m
es
f
r
o
m
t
h
e
w
a
y
w
e
co
m
p
u
ted
t
h
e
en
er
g
y
o
f
th
e
C
V
s
eg
m
e
n
ts
[
5
6
]
.
T
h
e
m
ai
n
to
o
ls
w
e
u
s
ed
ar
e
M
A
T
L
A
B
co
d
es
an
d
t
h
e
to
o
lb
o
x
P
r
aa
t
[
5
7
]
,
[
5
8
]
.
2
.
3
.
Co
m
p
uta
t
io
n o
f
lo
g
a
rit
h
m
ic
ener
g
y
cha
ra
ct
er
is
t
ics ba
s
ed
o
n DF
T
I
n
th
e
p
r
e
-
p
r
o
ce
s
s
i
n
g
p
h
ase,
w
e
d
iv
i
d
ed
th
e
s
p
ee
ch
s
i
g
n
al,
s
a
m
p
led
at
2
2
0
5
0
Hz,
in
to
ti
m
e
s
eg
m
en
ts
o
f
1
1
.
6
m
s
w
it
h
a
n
o
v
er
lap
o
f
9
.
6
m
s
.
T
h
er
ea
f
ter
w
e
a
p
p
lied
a
h
a
m
m
i
n
g
w
i
n
d
o
w
i
n
g
to
ea
c
h
s
eg
m
e
n
t
f
o
llo
w
ed
b
y
ze
r
o
-
p
ad
d
in
g
.
Fi
n
all
y
,
w
e
ca
lcu
lated
a
5
1
2
-
p
o
in
t
d
is
cr
ete
f
o
u
r
ier
tr
an
s
f
o
r
m
.
No
w
,
to
co
m
p
u
te
th
e
en
er
g
y
a
n
d
its
d
is
tr
ib
u
tio
n
f
o
r
ea
ch
s
y
llab
le
as
s
h
o
w
n
in
Fig
u
r
e
3
,
w
e
ch
o
s
e
s
i
x
s
p
ec
i
f
ic
b
a
n
d
s
o
f
f
r
eq
u
en
c
y
as
in
[
5
9
]
,
[
6
0
]
:
(
1
:
0
–
400
Hz
)
;
2
:
400
–
800
Hz
;
3
:
800
–
1200
Hz
;
4
:
1200
–
2000
Hz
;
5
:
2000
–
3500
Hz
6
:
3500
–
5000
Hz
)
.
Fig
u
r
e
3
.
B
lo
ck
d
iag
r
a
m
o
f
t
h
e
co
m
p
u
te
o
f
e
n
er
g
y
in
s
ix
b
a
n
d
s
T
h
ese
f
r
eq
u
e
n
c
y
b
a
n
d
s
co
r
r
esp
o
n
d
in
f
ac
t
to
ea
c
h
p
ar
t
o
f
t
h
e
v
o
ca
l
tr
ac
t.
Fro
m
t
h
e
s
m
o
o
th
ed
s
p
ec
tr
u
m
X(
n
,
s
)
(
t
h
e
s
ca
le
s
p
e
ctr
u
m
f
o
r
ea
ch
f
r
a
m
e
w
as
s
m
o
o
th
ed
b
y
a
2
0
p
o
in
t
m
o
v
in
g
a
v
er
ag
e
ta
k
e
n
alo
n
g
th
e
ti
m
e
in
d
e
x
n
)
,
w
e
ca
lcu
late
d
th
e
en
er
g
y
i
n
ea
ch
b
an
d
b
y
(
1
)
:
(
)
=
∑
10
l
og
10
|
(
,
)
|
2
(
1
)
w
h
er
e
t
h
e
b
an
d
r
ec
o
r
d
B
g
o
es
f
r
o
m
1
to
6
.
T
h
e
f
r
eq
u
en
c
y
in
d
ex
s
r
an
g
es
f
r
o
m
t
h
e
DF
T
(
Dis
cr
ete
Fo
u
r
ier
T
r
an
s
f
o
r
m
)
i
n
d
ices
r
ep
r
esen
ti
n
g
b
o
u
n
d
ar
ies
(
t
h
e
lo
w
er
a
n
d
u
p
p
er
)
f
o
r
ea
c
h
f
r
eq
u
e
n
c
y
b
an
d
.
Af
ter
th
at,
t
h
e
n
o
r
m
aliza
t
io
n
is
ap
p
lied
to
ea
ch
f
r
a
m
e,
ac
co
r
d
in
g
to
(
2
)
:
(
)
=
(
)
(
)
(
2
)
w
h
er
e
is
th
e
s
ta
n
d
ar
d
ized
b
a
n
d
en
er
g
y
B
i
n
t
h
e
f
r
a
m
e
n
,
(
n
)
is
t
h
e
g
en
er
al
e
n
er
g
y
in
t
h
e
f
r
a
m
e
n
an
d
(
)
is
th
e
b
an
d
e
n
er
g
y
B
i
n
th
e
f
r
a
m
e
n.
3.
T
H
E
R
E
CO
G
NI
T
I
O
N
M
O
DE
L
T
h
e
r
ec
o
g
n
itio
n
m
o
d
el
co
n
s
i
s
ts
o
f
t
w
o
s
tep
s
:
(
1
)
ex
tr
ac
ti
n
g
th
e
ac
o
u
s
tic
f
ea
tu
r
e
s
to
h
av
e
a
d
ata
s
et
(
2
)
ap
p
ly
in
g
t
h
e
c
lass
if
ica
tio
n
m
o
d
el.
T
h
e
d
ata
s
et
is
o
b
tai
n
ed
b
y
u
s
in
g
P
r
aa
t
a
n
d
M
A
T
L
A
B
co
d
es.
Fo
r
t
h
e
task
o
f
class
i
f
ica
tio
n
,
w
e
u
tili
z
ed
th
e
KNN
[
6
1
]
to
class
if
y
a
n
in
s
ta
n
ce
o
f
a
d
ata
s
et
i
n
to
an
e
m
o
tio
n
cla
s
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
E
mo
tio
n
r
ec
o
g
n
itio
n
fr
o
m
s
ylla
b
ic
u
n
its
u
s
in
g
…
(
A
b
d
ella
h
A
g
r
ima
)
5443
T
h
e
KNN
alg
o
r
ith
m
i
s
a
s
u
p
er
v
is
ed
lear
n
i
n
g
tec
h
n
iq
u
e;
it
ca
n
b
e
u
s
ed
f
o
r
b
o
th
class
if
i
ca
tio
n
an
d
r
eg
r
ess
io
n
.
T
o
m
ak
e
a
p
r
ed
ictio
n
,
th
e
KN
N
alg
o
r
it
h
m
w
ill
b
e
f
o
u
n
d
ed
o
n
t
h
e
en
t
ir
e
d
ataset.
I
n
d
ee
d
,
f
o
r
an
o
b
s
er
v
atio
n
,
w
h
ich
i
s
n
o
t
p
ar
t
o
f
th
e
d
ataset,
t
h
at
w
e
w
a
n
t
to
p
r
ed
ict,
th
e
alg
o
r
ith
m
w
i
l
l
s
ea
r
ch
f
o
r
th
e
K
in
s
ta
n
ce
s
o
f
th
e
d
ataset
clo
s
est
to
o
u
r
o
b
s
er
v
at
io
n
.
T
h
en
f
o
r
t
h
ese
K
n
ei
g
h
b
o
r
s
,
t
h
e
al
g
o
r
ith
m
i
s
b
ased
o
n
t
h
ei
r
o
u
tp
u
t
v
ar
iab
le
s
Y
to
ca
lcu
lat
e
th
e
v
al
u
e
o
f
t
h
e
v
ar
iab
le
Y
o
f
th
e
o
b
s
er
v
atio
n
th
at
w
e
w
a
n
t
to
p
r
ed
ict.
A
ls
o
i
f
K
-
NN
is
u
s
ed
f
o
r
th
e
clas
s
i
f
ic
atio
n
,
it
is
th
e
m
o
d
e
o
f
th
e
v
ar
iab
les
Y
o
f
th
e
K
clo
s
est
o
b
s
er
v
atio
n
s
w
h
ic
h
w
ill
b
e
u
s
ed
f
o
r
th
e
p
r
ed
ictio
n
.
3
.
1
.
Alg
o
rit
h
m
ic
co
m
po
s
it
io
n
W
e
ca
n
r
ep
r
esen
t th
e
o
p
er
atio
n
s
o
f
K
NN
b
y
w
r
iti
n
g
th
e
f
o
ll
o
w
i
n
g
p
s
e
u
d
o
-
co
d
e:
A
l
g
o
r
ith
m
1
:
Star
t
A
l
g
o
r
it
h
m
I
n
p
u
t d
ata:
a)
A
d
ata
s
et
D.
b)
A
f
u
n
ctio
n
f
o
r
d
ef
i
n
i
n
g
t
h
e
d
is
tan
ce
d
.
c)
An
i
n
teg
er
K
Fo
r
a
n
e
w
o
b
s
er
v
at
io
n
X
f
o
r
w
h
ich
w
e
w
an
t to
p
r
ed
ict
its
o
u
t
p
u
t v
ar
iab
le
Y
d
o
:
Step
1
:
C
alcu
late
all
t
h
e
d
is
ta
n
ce
s
o
f
th
i
s
o
b
s
er
v
atio
n
X
w
it
h
th
e
o
th
er
o
b
s
er
v
atio
n
s
o
f
t
h
e
d
ata
s
et
D.
Step
2
:
R
etain
th
e
K
o
b
s
er
v
ati
o
n
s
f
r
o
m
t
h
e
d
ata
s
et
D
clo
s
est
to
X
u
s
in
g
t
h
e
f
u
n
c
tio
n
o
f
c
alcu
lati
n
g
th
e
d
is
tan
ce
d
.
Step
3
:
T
ak
e
th
e
v
alu
e
s
o
f
Y
f
r
o
m
t
h
e
K
o
b
s
er
v
atio
n
s
r
etai
n
ed
:
a)
I
f
w
e
p
er
f
o
r
m
a
r
eg
r
ess
io
n
,
ca
l
cu
late
t
h
e
m
ea
n
(
o
r
m
ed
ian
)
o
f
Y
r
etain
ed
.
b)
I
f
w
e
p
er
f
o
r
m
a
class
if
ica
tio
n
,
ca
lcu
late
t
h
e
m
o
d
e
o
f
Y
r
etai
n
ed
(
th
is
is
o
u
r
ca
s
e)
.
c)
R
etu
r
n
th
e
v
al
u
e
ca
lcu
lated
in
s
tep
3
as th
e
v
al
u
e
t
h
at
w
as p
r
ed
icted
b
y
K
-
NN
f
o
r
o
b
s
er
v
atio
n
X.
E
n
d
K
-
NN
n
ee
d
s
a
d
is
tan
ce
ca
lcu
l
atio
n
f
u
n
ct
io
n
b
et
w
ee
n
t
w
o
o
b
s
er
v
atio
n
s
.
I
n
o
u
r
ca
s
e,
w
e
h
a
v
e
co
n
ti
n
u
o
u
s
d
ata;
h
en
ce
t
h
e
E
u
c
lid
ea
n
d
is
ta
n
ce
i
s
a
g
o
o
d
ca
n
d
id
ate.
3
.
2
.
E
uclid
ea
n dis
t
a
nce
I
t
is
a
d
is
tan
ce
th
at
co
m
p
u
te
s
th
e
s
q
u
ar
e
r
o
o
t
o
f
th
e
s
u
m
o
f
t
h
e
s
q
u
ar
e
d
if
f
er
en
ce
s
b
e
t
w
ee
n
t
h
e
co
o
r
d
in
ates o
f
t
w
o
p
o
in
t
s
:
d
(
x
,
y
)
=
√
∑
=
1
(
−
)
2
(
3
)
w
h
er
e
x
=
(
)
an
d
y
=
(
)
;
j=1
...
n
Fo
r
all
o
u
r
ex
p
er
i
m
e
n
ts
,
w
e
u
s
ed
th
e
f
r
ee
s
o
f
t
w
ar
e
ST
A
T
I
ST
I
C
A
[
6
2
]
w
h
ic
h
is
a
s
et
o
f
d
ata
m
i
n
i
n
g
to
o
ls
allo
w
in
g
t
h
e
p
r
o
ce
s
s
i
n
g
an
d
s
elec
t
io
n
o
f
t
h
e
p
ar
a
m
et
er
s
an
d
p
r
o
p
o
s
in
g
d
i
f
f
er
en
t
le
ar
n
in
g
al
g
o
r
ith
m
s
.
T
h
is
s
o
f
t
w
ar
e
i
s
cu
r
r
en
tl
y
in
c
r
ea
s
in
g
l
y
u
s
ed
in
t
h
e
p
atter
n
r
ec
o
g
n
itio
n
co
m
m
u
n
it
y
.
I
t
i
n
c
lu
d
es
m
an
y
k
n
o
w
n
alg
o
r
ith
m
s
s
u
c
h
as K
NN,
SV
M,
d
ec
is
io
n
tr
ee
s
(
J
4
8
)
,
as
w
el
l a
s
Me
ta
alg
o
r
it
h
m
s
.
ST
A
T
I
S
T
I
C
A
K
NN
i
s
a
m
e
m
o
r
y
-
b
a
s
ed
m
o
d
el
c
h
ar
ac
ter
i
ze
d
b
y
a
b
u
n
ch
o
f
e
x
a
m
p
les
(
o
b
j
ec
ts
)
f
o
r
w
h
ic
h
t
h
e
r
esu
lts
ar
e
k
n
o
w
n
(
i.e
.
,
th
e
ex
a
m
p
les
ar
e
lab
e
led
)
.
I
n
KNN
th
e
i
n
d
ep
en
d
e
n
t
a
n
d
d
ep
en
d
en
t
v
ar
iab
les
ca
n
b
e
eith
er
ca
teg
o
r
ical
o
r
co
n
tin
u
o
u
s
.
T
h
e
p
r
o
b
le
m
is
th
e
r
eg
r
e
s
s
io
n
f
o
r
co
n
ti
n
u
o
u
s
d
ep
en
d
en
t
v
ar
iab
les,
o
th
er
w
is
e,
t
h
e
p
r
o
b
le
m
is
t
h
e
class
if
ica
tio
n
.
H
en
ce
,
KNN
in
ST
A
T
I
STI
C
A
c
a
n
h
a
n
d
le
b
o
th
class
i
f
icatio
n
an
d
r
eg
r
e
s
s
io
n
p
r
o
b
lem
s
.
I
n
th
e
e
v
e
n
t
th
a
t
we
h
av
e
a
n
o
th
er
m
o
d
el
(
o
b
j
ec
t
)
,
w
e
w
o
u
ld
li
k
e
to
ap
p
r
o
x
im
a
te
th
e
o
u
tco
m
e
d
ep
en
d
en
t
o
n
t
h
e
KNN
e
x
a
m
p
le
s
.
T
o
s
ettle
o
n
th
e
ch
o
ice
KNN
s
h
o
u
ld
d
is
co
v
er
K
m
o
d
el
s
t
h
at
ar
e
n
ea
r
est
in
d
i
s
tan
ce
to
o
u
r
n
e
w
m
o
d
el
(
o
b
j
ec
t)
.
KNN
p
r
ed
ictio
n
s
d
ep
en
d
o
n
av
er
ag
i
n
g
th
e
r
esu
lt
s
o
f
th
e
K
-
Nea
r
est
-
Nei
g
h
b
o
r
f
o
r
th
e
r
eg
r
ess
io
n
p
r
o
b
le
m
s
.
Fo
r
th
e
class
i
f
ica
tio
n
p
r
o
b
le
m
s
,
KNN
u
tili
ze
s
th
e
v
o
te
d
o
m
in
a
n
t
p
ar
t
r
u
le.
T
h
e
v
alu
e
o
f
K
s
tr
o
n
g
l
y
i
m
p
ac
ts
th
e
p
r
ed
ictio
n
a
cc
u
r
ac
y
.
T
o
f
in
d
th
e
o
p
ti
m
a
l
v
alu
e
f
o
r
K
w
e
ca
n
u
tili
ze
t
h
e
cr
o
s
s
-
v
alid
atio
n
al
g
o
r
ith
m
i
n
ST
A
T
I
S
T
I
C
A
.
4.
E
XP
E
R
I
M
E
NT
A
L
RE
SUL
T
S
W
e
h
av
e
r
u
n
th
e
K
NN
al
g
o
r
ith
m
s
ev
er
al
ti
m
es,
ea
c
h
t
i
m
e
ai
m
i
n
g
f
o
r
a
d
i
f
f
er
e
n
t
g
o
al.
T
h
e
class
i
f
icatio
n
s
w
e
m
ad
e
w
er
e
ac
co
r
d
in
g
l
y
b
i
n
ar
y
o
r
m
u
l
tip
le.
A
ll
e
x
p
er
i
m
en
t
s
w
er
e
p
er
f
o
r
m
ed
o
n
d
ata
s
e
ts
co
llected
f
r
o
m
s
y
llab
ic
u
n
it
s
: /
b
a/,
/d
u
/,
/k
i/,
/ta/.
T
h
er
e
ar
e
f
o
u
r
class
i
f
icat
io
n
tas
k
s
p
r
ese
n
te
d
in
th
i
s
w
o
r
k
:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
6
,
Dec
em
b
er
2021
:
5
4
3
8
-
5
4
4
9
5444
a.
W
e
test
ed
th
e
ab
ili
t
y
o
f
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
to
d
etec
t
ea
ch
o
f
th
e
s
t
u
d
ied
e
m
o
tio
n
s
.
T
h
e
class
i
f
icatio
n
i
n
th
at
ca
s
e
is
b
in
ar
y
.
T
h
e
tar
g
e
ted
e
m
o
tio
n
(
N
:
n
e
u
tr
al,
H
:
j
o
y
,
S:
s
ad
n
es
s
,
A
:
a
n
g
er
)
w
a
s
lab
elled
b
y
i
ts
n
a
m
e
a
n
d
th
e
o
t
h
er
s
b
y
O
(
o
th
er
s
)
.
b.
Sti
ll
f
r
o
m
t
h
e
b
in
ar
y
cla
s
s
i
f
ica
tio
n
p
er
s
p
ec
tiv
e,
w
e
tr
ied
to
s
e
e
to
w
h
a
t e
x
ten
t o
u
r
m
o
d
el
i
s
ab
le
to
s
ep
ar
ate
a
m
o
n
g
p
o
s
iti
v
e
an
d
n
e
g
ati
v
e
e
m
o
tio
n
s
.
c.
W
ith
in
ea
ch
g
r
o
u
p
o
f
e
m
o
ti
o
n
(
p
o
s
itiv
e
an
d
n
e
g
ati
v
e)
,
w
e
test
ed
i
f
th
e
p
r
o
p
o
s
ed
f
ea
tu
r
e
v
ec
to
r
is
a
s
atis
f
ac
to
r
y
to
o
l
to
d
is
tin
g
u
is
h
b
et
w
ee
n
t
h
e
m
(
ac
co
r
d
in
g
t
o
[
6
3
]
,
p
o
s
itiv
e
e
m
o
t
io
n
s
i
n
clu
d
e
(
j
o
y
an
d
n
eu
tr
al)
a
n
d
n
eg
a
tiv
e
e
m
o
tio
n
s
in
cl
u
d
e
(
s
ad
n
es
s
an
d
an
g
er
)
)
.
d.
A
t la
s
t,
a
m
u
ltip
le
cla
s
s
i
f
icatio
n
is
p
er
f
o
r
m
ed
to
ev
al
u
ate
t
h
e
w
h
o
le
s
y
s
te
m
.
Fro
m
th
e
s
e
ex
p
er
i
m
e
n
t
s
,
r
esu
lts
ar
e
p
r
esen
ted
b
elo
w
.
A
s
s
h
o
w
n
i
n
Fi
g
u
r
e
4
,
th
e
a
n
al
y
s
is
o
f
th
e
s
y
llab
ic
u
n
it
/b
a/
u
s
in
g
t
h
e
p
r
o
p
o
s
ed
s
et
o
f
f
ea
tu
r
es
g
i
v
es
h
ig
h
ac
c
u
r
ac
y
p
er
ce
n
ta
g
e
o
f
d
e
tectio
n
.
I
n
d
ee
d
,
w
e
o
b
tain
ed
f
o
r
ex
a
m
p
le
6
0
.
2
0
%
f
o
r
all
e
m
o
tio
n
s
an
d
7
3
.
6
3
%
to
d
is
tin
g
u
i
s
h
b
et
w
ee
n
p
o
s
i
tiv
e
a
n
d
n
eg
a
tiv
e
e
m
o
tio
n
s
.
I
n
th
e
s
a
m
e
w
a
y
,
t
h
e
r
ates
v
ar
y
r
e
s
p
ec
tiv
el
y
f
r
o
m
8
6
.
2
1
%
to
7
8
.
9
5
%
f
o
r
p
o
s
itiv
e
e
m
o
tio
n
s
(
n
eu
tr
al
v
s
j
o
y
)
an
d
n
e
g
ati
v
e
e
m
o
t
io
n
s
(
an
g
er
v
s
s
ad
n
ess
)
.
T
h
e
KN
N
alg
o
r
ith
m
r
ec
o
g
n
ized
a
n
e
u
tr
al
s
tate
w
it
h
m
o
r
e
th
an
8
9
.
5
5
%,
s
ad
n
es
s
w
it
h
8
5
.
0
7
%,
an
g
er
w
it
h
7
9
.
1
0
%,
an
d
j
o
y
w
ith
7
6
.
1
2
%.
Fig
u
r
e
4
.
An
al
y
s
is
o
f
th
e
s
y
l
la
b
le
/b
a/
I
n
th
e
s
a
m
e
w
a
y
,
a
n
an
al
y
s
is
p
er
f
o
r
m
ed
o
n
th
e
C
V
/d
u
/
s
h
o
w
s
r
es
u
lt
s
w
it
h
r
ates
t
h
at
ca
n
g
o
u
p
to
9
4
.
9
5
%
to
d
etec
t
n
eu
tr
al
e
m
o
t
io
n
.
W
e
o
b
tain
ed
9
1
.
4
1
%
to
r
ec
o
g
n
ize
j
o
y
,
8
3
.
8
4
%
f
o
r
a
n
g
er
,
an
d
7
9
.
8
0
%
f
o
r
s
ad
n
es
s
.
Fo
r
a
n
al
y
s
is
b
et
w
ee
n
p
o
s
itiv
e
an
d
n
e
g
ati
v
e
e
m
o
tio
n
s
,
t
h
e
r
ate
r
ea
ch
ed
8
4
.
3
4
%,
an
d
9
1
.
5
7
%,
7
0
.
1
8
%
f
o
r
p
o
s
itiv
e
e
m
o
tio
n
s
(
n
e
u
tr
al
v
s
j
o
y
)
,
an
d
n
e
g
ati
v
e
e
m
o
t
io
n
s
(
an
g
er
v
s
s
ad
n
e
s
s
)
.
T
h
e
g
l
o
b
al
class
if
icat
io
n
in
cl
u
d
in
g
all
e
m
o
t
io
n
s
r
ea
c
h
e
d
a
r
ate
o
f
6
4
.
6
5
%
as sh
o
w
n
i
n
Fi
g
u
r
e
5
.
Fig
u
r
e
5
.
An
al
y
s
is
o
f
th
e
s
y
l
la
b
le
/d
u
/
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
E
mo
tio
n
r
ec
o
g
n
itio
n
fr
o
m
s
ylla
b
ic
u
n
its
u
s
in
g
…
(
A
b
d
ella
h
A
g
r
ima
)
5445
Si
m
i
lar
l
y
,
F
ig
u
r
e
6
s
h
o
w
s
th
e
ac
cu
r
ac
ies
t
h
at
w
e
o
b
tain
e
d
f
o
r
th
e
C
V
/
k
i/.
T
h
e
KNN
alg
o
r
ith
m
r
ec
o
g
n
ized
j
o
y
w
it
h
m
o
r
e
t
h
a
n
9
3
.
2
0
%,
an
g
er
w
it
h
8
4
.
4
7
%,
8
1
.
5
5
%
f
o
r
th
e
n
eu
tr
al
s
t
ate,
an
d
7
8
.
1
5
%
f
o
r
s
ad
n
es
s
.
A
ls
o
,
t
h
e
b
i
n
ar
y
cla
s
s
if
icatio
n
b
et
w
ee
n
p
o
s
iti
v
e
a
n
d
n
e
g
ati
v
e
e
m
o
t
io
n
s
ac
h
ie
v
ed
a
r
ec
o
g
n
itio
n
r
ate
o
f
7
2
.
3
3
%.
Fu
r
th
er
m
o
r
e,
b
y
co
n
s
id
er
in
g
ea
ch
g
r
o
u
p
o
f
e
m
o
ti
o
n
s
(
p
o
s
iti
v
e
an
d
n
e
g
ati
v
e)
,
th
e
r
es
u
lts
r
ea
ch
e
d
8
7
.
6
2
%
f
o
r
th
e
p
o
s
itiv
e
e
m
o
tio
n
s
a
n
d
7
5
%
f
o
r
th
e
n
e
g
ativ
e
o
n
e
s
.
C
o
n
ce
r
n
i
n
g
th
e
c
la
s
s
i
f
icatio
n
o
f
al
l
e
m
o
tio
n
s
,
th
e
p
er
f
o
r
m
ed
r
ate
r
ea
ch
ed
its
m
in
i
m
u
m
v
al
u
e
o
f
5
2
.
4
3
%.
Fin
all
y
,
t
h
e
s
t
u
d
y
o
f
t
h
e
C
V
/t
a/
s
h
o
w
s
t
h
at:
n
e
u
tr
al
s
tate
p
r
e
s
en
t
s
a
h
i
g
h
r
ec
o
g
n
itio
n
r
ate
o
f
9
0
.
1
3
%,
f
o
llo
w
ed
b
y
a
n
g
er
w
it
h
8
4
.
5
7
%,
s
ad
n
e
s
s
w
it
h
7
8
.
1
5
%,
an
d
last
l
y
j
o
y
w
it
h
7
7
.
1
6
%.
Fo
r
n
eg
ati
v
e
v
s
p
o
s
it
iv
e
e
m
o
tio
n
s
,
t
h
e
ac
cu
r
ac
y
i
s
7
5
.
9
2
%.
W
ith
in
ea
ch
ca
teg
o
r
y
o
f
e
m
o
tio
n
(
p
o
s
itiv
e
an
d
n
e
g
ati
v
e)
,
th
e
ac
cu
r
ac
y
i
s
8
9
.
7
4
%
an
d
7
7
.
3
8
%
r
esp
ec
ti
v
el
y
.
T
h
e
r
ate
o
b
tain
ed
f
o
r
th
e
w
h
o
le
e
m
o
tio
n
s
is
6
1
.
7
3
%
w
h
ic
h
ca
n
b
e
co
n
s
id
er
ed
as a
n
ac
ce
p
tab
le
r
ate
as sh
o
w
n
i
n
Fi
g
u
r
e
7
.
Fig
u
r
e
6
.
An
al
y
s
is
o
f
th
e
s
y
l
la
b
le
/k
i/
Fig
u
r
e
7
.
An
al
y
s
is
o
f
th
e
s
y
l
la
b
le
/ta/
5.
DIS
CU
SS
I
O
N
Ou
r
s
t
u
d
y
ai
m
s
to
p
r
o
v
id
e
a
n
au
to
m
a
tic
e
m
o
tio
n
r
ec
o
g
n
iti
o
n
m
o
d
el
w
i
th
a
r
ed
u
ce
d
s
et
o
f
ac
o
u
s
tic
f
ea
t
u
r
es.
Me
as
u
r
e
m
en
t
s
w
er
e
ca
r
r
ied
o
u
t
u
s
in
g
th
e
M
A
DE
D
d
atab
ase
f
r
o
m
w
h
ic
h
w
e
ex
tr
ac
ted
f
o
u
r
p
lo
s
i
v
e
co
n
s
o
n
a
n
t
/b
/,
/d
/,
/
k
/,
/t/
a
s
s
o
ciate
d
w
i
th
t
h
e
t
h
r
ee
v
o
w
el
s
/a
/,
/u
/
a
n
d
/i/.
T
h
ese
c
h
o
ices
ar
e
b
ased
o
n
p
r
ev
io
u
s
w
o
r
k
s
[
6
4
]
,
w
h
er
e
t
h
e
a
u
t
h
o
r
s
p
r
o
v
ed
th
at
th
e
en
er
g
y
a
n
d
it
s
d
is
tr
ib
u
tio
n
in
s
p
ec
i
f
ic
b
a
n
d
s
p
la
y
an
i
m
p
o
r
ta
n
t
r
o
le
in
ch
ar
ac
ter
is
in
g
a
r
ab
ic
p
lo
s
iv
e
co
n
s
o
n
a
n
ts
.
T
h
e
r
esu
lts
w
e
o
b
tain
ed
ar
e
q
u
ite
s
atis
f
a
cto
r
y
co
m
p
ar
i
n
g
to
p
r
ev
io
u
s
w
o
r
k
s
as s
h
o
w
n
i
n
T
ab
le
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
6
,
Dec
em
b
er
2021
:
5
4
3
8
-
5
4
4
9
5446
I
t
s
h
o
u
ld
b
e
p
o
in
ted
o
u
t
t
h
at
o
u
r
s
t
u
d
y
r
aised
m
an
y
q
u
e
s
tio
n
s
.
I
n
d
ee
d
,
it
is
s
ee
n
f
r
o
m
th
e
r
esu
lt
s
t
h
at
co
n
s
o
n
a
n
t
/d
/
ac
h
iev
e
s
t
h
e
b
est
r
ates
i
n
al
m
o
s
t
all
ca
s
e
s
(
e
s
p
ec
iall
y
i
n
t
h
e
n
e
u
tr
al
ca
s
e
9
4
.
9
5
%)
w
h
ic
h
m
a
y
lead
u
s
to
th
i
n
k
t
h
at
p
lace
o
f
a
r
ticu
latio
n
o
f
th
e
co
n
s
o
n
a
n
t
h
a
s
a
r
o
le
to
p
lay
in
d
eter
m
i
n
i
n
g
th
e
e
m
o
tio
n
u
n
d
er
co
n
s
id
er
atio
n
.
Mo
r
eo
v
er
,
s
y
ll
ab
les
ass
o
ciate
d
w
it
h
t
h
e
s
a
m
e
v
o
w
el
(
/b
a/
an
d
/ta/)
s
ee
m
t
o
p
r
esen
t
al
m
o
s
t
t
h
e
s
a
m
e
r
es
u
lt
s
.
B
u
t a
s
w
e
co
m
e
to
in
v
e
s
ti
g
ate
m
o
r
e
ca
r
ef
u
ll
y
t
h
e
r
ec
o
g
n
i
tio
n
r
ates,
w
e
ca
n
s
e
e
th
at:
a.
Neg
ati
v
e
e
m
o
tio
n
s
p
r
esen
t
t
h
e
b
est
r
ates
(
7
8
.
9
5
%
f
o
r
/b
a/
an
d
7
7
.
3
8
%
f
o
r
/ta/
w
h
ile
f
o
r
/
d
u
/
7
0
.
1
8
%
an
d
f
o
r
/k
i/ 7
5
%)
b.
W
h
en
th
e
v
o
w
e
l
/a/
is
ass
o
ciate
d
w
it
h
th
e
p
lo
s
iv
e
/b
/,
s
ad
n
ess
is
m
o
r
e
r
ec
o
g
n
ized
t
h
an
a
n
g
er
.
T
h
e
o
p
p
o
s
ite
o
cc
u
r
s
w
h
e
n
/a/
i
s
as
s
o
ciate
d
w
it
h
/t/.
A
s
li
g
h
t
co
m
p
ar
i
s
o
n
b
et
w
ee
n
s
y
llab
les
/
k
i/
an
d
/d
u
/
s
h
o
w
s
th
at
f
o
r
b
o
th
th
e
j
o
y
p
r
esen
t
s
t
h
e
b
est
r
ec
o
g
n
itio
n
r
ates
(
9
3
.
2
0
%
r
e
s
p
.
9
1
.
4
1
%).
B
u
t
d
if
f
er
e
n
ce
s
o
cc
u
r
in
th
e
n
e
u
tr
al
an
d
m
u
ltip
le
clas
s
i
f
icatio
n
ca
s
es.
T
h
ese
r
ates
estab
lis
h
i
n
f
ac
t
h
o
w
f
ar
o
b
j
ec
ts
f
r
o
m
t
h
e
e
m
o
tio
n
a
l
r
ep
r
esen
ta
tio
n
w
e
p
r
o
p
o
s
e
ar
e
cl
o
s
e
to
ea
ch
o
th
er
.
I
n
d
ee
d
,
th
e
e
x
p
lo
r
ato
r
y
n
at
u
r
e
o
f
o
u
r
s
tu
d
y
h
a
s
d
ictated
th
e
ch
o
ice
o
f
KNN
al
g
o
r
ith
m
r
ath
er
t
h
a
n
SVM
o
r
ar
tif
icial
n
eu
r
o
n
a
l
n
et
w
o
r
k
(
A
NN)
.
in
th
e
cla
s
s
i
f
icat
io
n
task
.
O
u
r
m
ain
co
n
ce
r
n
is
to
estab
lis
h
to
h
o
w
ex
ten
t th
e
f
ea
t
u
r
es v
ec
to
r
s
u
cc
ee
d
s
to
ev
alu
ate
s
i
m
ilar
ities
b
e
t
w
ee
n
t
h
e
s
a
m
e
e
m
o
tio
n
s
.
6.
CO
NCLU
SI
O
N
T
h
is
w
o
r
k
g
i
v
es
a
g
o
o
d
g
r
o
u
n
d
in
g
i
n
m
o
d
eli
n
g
e
m
o
tio
n
with
ac
o
u
s
tic
f
ea
tu
r
es.
T
h
e
m
e
t
h
o
d
g
i
v
e
n
h
er
e
u
s
es
e
n
er
g
y
a
n
d
its
d
is
tr
i
b
u
tio
n
i
n
s
ix
b
an
d
s
a
s
a
p
r
in
ci
p
al
to
o
l
f
o
r
d
is
tin
g
u
is
h
i
n
g
b
etw
ee
n
t
h
e
f
o
u
r
b
asi
c
e
m
o
tio
n
s
:
n
e
u
tr
al,
s
ad
n
es
s
,
j
o
y
,
a
n
d
an
g
er
.
T
h
e
cla
s
s
ica
l
K
NN
al
g
o
r
ith
m
is
u
s
ed
to
p
er
f
o
r
m
t
h
e
c
lass
if
ica
tio
n
task
.
I
n
s
o
m
e
ca
s
es,
t
h
e
r
e
s
u
l
ts
w
er
e
co
n
c
lu
s
i
v
e
b
u
t
n
o
t
e
x
h
a
u
s
tiv
e.
T
h
is
s
tu
d
y
ca
n
b
e
ex
te
n
d
ed
in
f
u
t
u
r
e
w
o
r
k
s
to
r
ic
h
er
co
r
p
o
r
a
w
it
h
d
if
f
er
en
t
u
t
ter
an
ce
r
ep
r
ese
n
t
atio
n
s
in
d
if
f
er
en
t
la
n
g
u
a
g
es
an
d
w
it
h
d
if
f
er
en
t
alg
o
r
ith
m
s
l
ik
e
n
eu
r
al
n
et
w
o
r
k
s
a
n
d
s
u
p
p
o
r
t v
ec
to
r
m
ac
h
in
e
alg
o
r
ith
m
s
.
RE
F
E
R
E
NC
E
S
[1
]
L
.
A
.
P
.
G
a
sp
a
r,
S
.
O.
C.
M
o
r
a
les
,
a
n
d
F
.
T
.
Ro
m
e
ro
,
“
M
u
l
ti
m
o
d
a
l
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
w
it
h
Ev
o
lu
ti
o
n
a
ry
Co
m
p
u
tatio
n
f
o
r
Hu
m
a
n
-
Ro
b
o
t
In
tera
c
ti
o
n
,
”
Ex
p
e
rt
S
y
ste
m
s
w
it
h
A
p
p
li
c
a
ti
o
n
s,
v
o
l.
6
6
,
p
p
.
4
2
-
6
1
,
2
0
1
6
,
d
o
i:
1
0
.
1
0
1
6
/j
.
e
sw
a
.
2
0
1
6
.
0
8
.
0
4
7
.
[2
]
L.
F
.
Ch
e
n
,
Z
.
T
.
L
iu
,
M
.
W
u
,
M
.
Din
g
,
F
.
Y.
Do
n
g
,
a
n
d
K.
Hiro
ta
,
“
Em
o
ti
o
n
-
A
g
e
-
Ge
n
d
e
r
-
Na
ti
o
n
a
li
ty
Ba
se
d
In
ten
ti
o
n
Un
d
e
rsta
n
d
i
n
g
in
Hu
m
a
n
-
Ro
b
o
t
In
tera
c
ti
o
n
Us
in
g
T
w
o
-
La
y
e
r
F
u
z
z
y
S
u
p
p
o
rt
V
e
c
to
r
Re
g
re
ss
io
n
,
”
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
S
o
c
i
a
l
R
o
b
o
ti
c
s
,
v
o
l.
7
,
n
o
.
5
,
p
p
.
7
0
9
-
7
2
9
,
2
0
1
5
,
d
o
i:
1
0
.
1
0
0
7
/s1
2
3
6
9
-
0
1
5
-
0
2
9
0
-
2
[3
]
A
.
F
.
Ca
b
a
ll
e
ro
e
t
a
l
.,
“
S
m
a
rt
En
v
iro
n
m
e
n
t
A
rc
h
it
e
c
tu
re
f
o
r
E
m
o
ti
o
n
De
tec
ti
o
n
a
n
d
Re
g
u
latio
n
,
”
J
o
u
rn
a
l
o
f
Bi
o
me
d
ica
l
In
f
o
rm
a
ti
c
s
,
v
o
l
.
6
4
,
p
p
.
5
5
-
7
3
,
2
0
1
6
,
d
o
i:
1
0
.
1
0
1
6
/
j.
j
b
i.
2
0
1
6
.
0
9
.
0
1
5
.
[4
]
M
.
Eg
g
e
r
,
M
.
L
e
y
,
a
n
d
S
.
Ha
n
k
e
,
“
E
m
o
ti
o
n
Re
c
o
g
n
it
io
n
f
ro
m
P
h
y
sio
lo
g
ica
l
S
ig
n
a
l
A
n
a
l
y
si
s:
A
R
e
v
ie
w
,
”
El
e
c
tro
n
ic No
tes
i
n
T
h
e
o
re
ti
c
a
l
C
o
mp
u
ter
S
c
ien
c
e
,
v
o
l.
3
4
3
,
p
p
.
3
5
-
5
5
,
2
0
1
9
,
d
o
i:
1
0
.
1
0
1
6
/j
.
e
n
tcs
.
2
0
1
9
.
0
4
.
0
0
9
.
[5
]
H.
Bo
u
b
e
n
n
a
a
n
d
D.
L
e
e
,
“
I
m
a
g
e
-
Ba
se
d
E
m
o
ti
o
n
Re
c
o
g
n
it
i
o
n
Us
in
g
Ev
o
lu
ti
o
n
a
ry
A
l
g
o
rit
h
m
s,
”
Bi
o
lo
g
ica
ll
y
In
sp
ire
d
C
o
g
n
it
ive
Arc
h
it
e
c
tu
re
s
,
v
o
l.
2
4
,
p
p
.
7
0
-
7
6
,
2
0
1
8
,
d
o
i:
1
0
.
1
0
1
6
/j
.
b
ica
.
2
0
1
8
.
0
4
.
0
0
8
.
[6
]
A
.
R
a
h
e
e
l
,
M
.
M
a
ji
d
,
M
.
A
ln
o
w
a
m
i,
a
n
d
S
.
M
.
A
n
w
a
r
,
“
P
h
y
sio
lo
g
ica
l
S
e
n
so
rs
Ba
se
d
E
m
o
ti
o
n
Re
c
o
g
n
it
io
n
W
h
il
e
Ex
p
e
rien
c
in
g
T
a
c
ti
le E
n
h
a
n
c
e
d
M
u
l
ti
m
e
d
ia,
”
S
e
n
so
rs
,
v
o
l.
2
0
,
n
o
.
1
4
,
2
0
2
0
,
A
rt.
n
o
.
4
0
3
7
,
d
o
i:
1
0
.
3
3
9
0
/s2
0
1
4
4
0
3
7
[7
]
W
.
M
e
ll
o
u
k
a
n
d
W
.
Ha
n
d
o
u
z
i,
“
F
a
c
ial
E
m
o
ti
o
n
Re
c
o
g
n
it
i
o
n
Us
in
g
De
e
p
L
e
a
rn
in
g
:
Re
v
ie
w
a
n
d
In
sig
h
ts,
”
Pro
c
e
d
ia
C
o
mp
u
ter
S
c
ien
c
e
,
v
o
l.
1
7
5
,
p
p
.
6
8
9
-
6
9
4
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
1
6
/j
.
p
ro
c
s.
2
0
2
0
.
0
7
.
1
0
1
.
[8
]
J.
W
u
,
S
.
G
u
o
,
H.
Hu
a
n
g
,
W
.
L
i
u
,
a
n
d
Y.
X
ian
g
,
“
I
n
f
o
rm
a
ti
o
n
a
n
d
Co
m
m
u
n
ica
ti
o
n
s
T
e
c
h
n
o
lo
g
ies
f
o
r
S
u
sta
in
a
b
le
De
v
e
lo
p
m
e
n
t
G
o
a
ls:
S
tate
-
of
-
th
e
-
A
rt,
Ne
e
d
s
a
n
d
P
e
rsp
e
c
ti
v
e
s
,”
IEE
E
Co
mm
u
n
ica
ti
o
n
s
S
u
rv
e
y
s
a
n
d
T
u
to
ri
a
ls
,
v
o
l.
2
0
,
n
o
.
3
,
p
p
.
2
3
8
9
-
2
4
0
6
,
2
0
1
8
,
d
o
i:
1
0
.
1
1
0
9
/COM
S
T
.
2
0
1
8
.
2
8
1
2
3
0
1
.
[9
]
M
.
J.
W
e
st,
A
.
J.
A
n
g
w
in
,
D.
A
.
Co
p
lan
d
,
W
.
L
.
A
rn
o
tt
,
a
n
d
N.
L
.
Ne
lso
n
,
“
Cr
o
ss
-
M
o
d
a
l
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
a
n
d
A
u
ti
s
m
-
li
k
e
T
ra
it
s
in
Ty
p
i
c
a
ll
y
De
v
e
lo
p
in
g
Ch
il
d
re
n
,
”
J
o
u
r
n
a
l
o
f
Exp
e
rime
n
ta
l
Ch
il
d
Psy
c
h
o
l
o
g
y
,
v
o
l.
1
9
1
,
2
0
2
0
,
A
rt.
n
o
.
1
0
4
7
3
7
,
d
o
i:
1
0
.
1
0
1
6
/j
.
jec
p
.
2
0
1
9
.
1
0
4
7
3
7
.
[1
0
]
M
.
Ja
n
ss
e
n
s
e
t
a
l.
,
“
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
i
n
P
sy
c
h
o
sis:
No
Ev
id
e
n
c
e
f
o
r
a
n
A
ss
o
c
iatio
n
w
it
h
Re
a
l
W
o
rld
S
o
c
ial
F
u
n
c
ti
o
n
i
n
g
,
”
S
c
h
izo
p
h
re
n
i
a
Res
e
a
rc
h
,
v
o
l.
1
4
2
,
n
o
.
1
-
3
,
p
p
.
1
1
6
-
1
2
1
,
2
0
1
2
,
d
o
i:
1
0
.
1
0
1
6
/j
.
sc
h
re
s.
2
0
1
2
.
1
0
.
0
0
3
.
[1
1
]
R.
Na
k
a
tsu
,
A
.
S
o
lo
m
id
e
s,
a
n
d
N.
T
o
sa
,
“
E
m
o
ti
o
n
Re
c
o
g
n
it
io
n
a
n
d
Its
A
p
p
li
c
a
ti
o
n
t
o
Co
m
p
u
t
e
r
A
g
e
n
ts
w
it
h
S
p
o
n
tan
e
o
u
s
In
tera
c
ti
v
e
Ca
p
a
b
il
it
ies
,
”
Kn
o
wle
d
g
e
-
Ba
se
d
S
y
st
e
ms
,
v
o
l.
1
3
,
n
o
.
7
-
8
,
p
p
.
4
9
7
-
5
0
4
,
2
0
0
0
,
d
o
i:
1
0
.
1
1
4
5
/3
1
9
4
6
3
.
3
1
9
6
4
1
.
[1
2
]
S
.
T
iv
a
tan
sa
k
u
l,
M
.
Oh
k
u
ra
,
S
.
P
u
a
n
g
p
o
n
ti
p
,
a
n
d
T
.
A
c
h
a
lak
u
l
,
“
Em
o
ti
o
n
a
l
He
a
lt
h
c
a
re
S
y
ste
m
:
E
m
o
ti
o
n
De
tec
ti
o
n
b
y
F
a
c
ial
Ex
p
re
ss
io
n
s
Us
in
g
J
a
p
a
n
e
se
Da
tab
a
s
e
,
”
2
0
1
4
6
t
h
Co
mp
u
ter
S
c
ien
c
e
a
n
d
El
e
c
tro
n
ic
En
g
i
n
e
e
rin
g
Co
n
fer
e
n
c
e
(
CEE
C)
,
2
0
1
4
,
p
p
.
4
1
-
46
,
d
o
i:
1
0
.
1
1
0
9
/C
EE
C.
2
0
1
4
.
6
9
5
8
5
5
2
.
[1
3
]
R.
L
.
M
a
n
d
ry
k
a
n
d
M
.
S
.
A
tk
in
s,
“
A
F
u
z
z
y
P
h
y
sio
lo
g
ica
l
A
p
p
ro
a
c
h
f
o
r
Co
n
ti
n
u
o
u
sly
M
o
d
e
li
n
g
Em
o
ti
o
n
d
u
rin
g
In
tera
c
ti
o
n
w
it
h
P
lay
T
e
c
h
n
o
lo
g
ies
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Hu
m
a
n
-
C
o
mp
u
ter
S
tu
d
ies
,
v
o
l.
6
5
,
n
o
.
4
,
pp
.
3
2
9
-
3
4
7
,
2
0
0
7
,
d
o
i:
1
0
.
1
0
1
6
/
j
.
ij
h
c
s.2
0
0
6
.
1
1
.
0
1
1
.
[1
4
]
D.
J
.
L
it
m
a
n
a
n
d
K.
F
o
rb
e
s
-
Ril
e
y
,
“
P
re
d
icti
n
g
stu
d
e
n
t
e
m
o
ti
o
n
s
i
n
c
o
m
p
u
ter
-
h
u
m
a
n
tu
to
ri
n
g
d
ialo
g
u
e
s
,
”
Pro
c
e
e
d
in
g
s
o
f
t
h
e
4
2
n
d
A
n
n
u
a
l
M
e
e
ti
n
g
o
f
t
h
e
Asso
c
ia
t
io
n
f
o
r
Co
mp
u
ta
ti
o
n
a
l
L
in
g
u
ist
ics
(
ACL
)
,
Ba
rc
e
lo
n
a
,
S
p
a
in
,
2
0
0
4
,
p
p
.
3
5
1
-
3
5
8
,
d
o
i:
1
0
.
3
1
1
5
/
1
2
1
8
9
5
5
.
1
2
1
9
0
0
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
E
mo
tio
n
r
ec
o
g
n
itio
n
fr
o
m
s
ylla
b
ic
u
n
its
u
s
in
g
…
(
A
b
d
ella
h
A
g
r
ima
)
5447
[1
5
]
M
.
Bo
jan
ić,
V
.
De
li
c
,
a
n
d
A
.
Ka
p
o
v
,
“
Ca
ll
R
e
d
istri
b
u
t
io
n
fo
r
a
Ca
ll
Ce
n
ter
B
a
se
d
o
n
S
p
e
e
c
h
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
,
”
Ap
p
li
e
d
S
c
ien
c
e
s
,
v
o
l.
1
0
,
n
o
.
1
3
,
2
0
2
0
,
A
rt.
n
o
.
4
6
5
3
,
d
o
i:
1
0
.
3
3
9
0
/a
p
p
1
0
1
3
4
6
5
3
.
[1
6
]
L
.
V
id
ra
sc
u
a
n
d
L
.
De
v
il
lers
,
“
Re
a
l
-
L
i
f
e
E
m
o
ti
o
n
Re
p
re
se
n
tati
o
n
a
n
d
De
tec
ti
o
n
in
Ca
ll
Ce
n
ters
Da
ta,
”
Af
fec
ti
v
e
Co
mp
u
t
in
g
a
n
d
In
telli
g
e
n
t
In
ter
a
c
ti
o
n
,
2
0
0
5
,
p
p
.
7
3
9
-
7
4
6
,
d
o
i:
1
0
.
1
0
0
7
/
1
1
5
7
3
5
4
8
_
9
5
.
[1
7
]
D.
M
o
rriso
n
,
R.
W
a
n
g
,
a
n
d
L
.
C.
De
S
il
v
a
,
“
En
se
m
b
le
M
e
th
o
d
s
f
o
r
S
p
o
k
e
n
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
in
Ca
ll
-
Ce
n
tres
,
”
S
p
e
e
c
h
Co
mm
u
n
ica
ti
o
n
,
v
o
l
.
4
9
,
n
o
.
2
,
p
p
.
9
8
-
1
1
2
,
2
0
0
7
,
d
o
i:
1
0
.
1
0
1
6
/j
.
sp
e
c
o
m
.
2
0
0
6
.
1
1
.
0
0
4
.
[1
8
]
C.
M
.
W
h
it
in
g
,
S
.
A
.
Ko
tz,
J.
Gro
ss
,
B.
L
.
G
io
rd
a
n
o
,
a
n
d
P
.
Be
li
n
,
“
T
h
e
P
e
rc
e
p
ti
o
n
o
f
Ca
rica
tu
re
d
Em
o
ti
o
n
in
V
o
ice
,
”
C
o
g
n
it
i
o
n
,
v
o
l.
2
0
0
,
2
0
2
0
,
A
rt.
n
o
.
1
0
4
2
4
9
,
d
o
i:
1
0
.
1
0
1
6
/j
.
c
o
g
n
it
io
n
.
2
0
2
0
.
1
0
4
2
4
9
.
[1
9
]
N
.
Ka
m
a
ru
d
d
in
,
A
.
W
a
h
a
b
,
a
n
d
C.
Qu
e
k
,
“
Cu
lt
u
ra
l
De
p
e
n
d
e
n
c
y
A
n
a
l
y
sis
f
o
r
Un
d
e
rsta
n
d
i
n
g
S
p
e
e
c
h
E
m
o
ti
o
n
,
”
Exp
e
rt S
y
ste
ms
wit
h
A
p
p
l
ica
ti
o
n
s
,
v
o
l.
3
9
,
n
o
.
5
,
p
p
.
5
1
1
5
-
5
1
3
3
,
2
0
1
2
,
d
o
i:
1
0
.
1
0
1
6
/
j.
e
sw
a
.
2
0
1
1
.
1
1
.
0
2
8
.
[2
0
]
G
.
D
a
v
id
,
J.
M
.
M
o
li
n
a
,
a
n
d
Z
.
Ca
ll
e
jas
,
“
Co
m
b
in
in
g
S
p
e
e
c
h
-
Ba
se
d
a
n
d
L
in
g
u
isti
c
Clas
si
f
iers
to
Re
c
o
g
n
ize
Em
o
ti
o
n
in
Us
e
r
S
p
o
k
e
n
Uttera
n
c
e
s,”
Ne
u
ro
c
o
mp
u
ti
n
g
,
v
o
l
.
3
2
6
-
3
2
7
,
p
p
.
1
3
2
-
1
4
0
,
2
0
1
9
,
d
o
i:
1
0
.
1
0
1
6
/j
.
n
e
u
c
o
m
.
2
0
1
7
.
0
1
.
1
2
0.
[2
1
]
S
.
L
Ca
stro
a
n
d
C.
F
L
im
a
,
“
Re
c
o
g
n
izin
g
e
m
o
ti
o
n
s
in
sp
o
k
e
n
lan
g
u
a
g
e
:
A
v
a
li
d
a
ted
se
t
o
f
P
o
rtu
g
u
e
se
se
n
ten
c
e
s
a
n
d
p
se
u
d
o
se
n
ten
c
e
s
f
o
r
re
se
a
r
c
h
o
n
e
m
o
ti
o
n
a
l
p
r
o
so
d
y
,”
Beh
a
v
io
r
Res
e
a
rc
h
M
e
th
o
d
s,
v
o
l.
4
2
,
n
o
.
1
,
p
p
.
7
4
-
8
1
,
2
0
1
0
,
d
o
i:
1
0
.
3
7
5
8
/BR
M
.
4
2
.
1
.
7
4
.
[2
2
]
S
.
M
o
h
it
,
M
.
T
u
,
V
.
Be
rish
a
,
C.
Ch
a
k
ra
b
a
rti
,
a
n
d
A
.
S
p
a
n
ias
,
“
A
r
ti
c
u
latio
n
Co
n
stra
i
n
e
d
L
e
a
rn
in
g
w
it
h
A
p
p
li
c
a
ti
o
n
to
S
p
e
e
c
h
Em
o
ti
o
n
Re
c
o
g
n
it
i
o
n
,”
EURA
S
IP
J
o
u
r
n
a
l
o
n
A
u
d
i
o
,
S
p
e
e
c
h
,
a
n
d
M
u
sic
Pro
c
e
ss
in
g
,
v
o
l.
2
0
1
9
,
n
o
.
1
,
2
0
1
9
,
A
rt.
n
o
.
1
4
,
d
o
i:
1
0
.
1
1
8
6
/s1
3
6
3
6
-
0
1
9
-
0
1
5
7
-
9.
[2
3
]
L
.
Ch
e
n
,
X
.
M
a
o
,
Y.
X
u
e
,
a
n
d
L
.
L
.
Ch
e
n
g
,
“
S
p
e
e
c
h
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
:
F
e
a
tu
re
s
a
n
d
Clas
sif
ic
a
ti
o
n
M
o
d
e
ls,
”
Dig
it
a
l
S
i
g
n
a
l
Pr
o
c
e
ss
in
g
,
v
o
l.
2
2
,
n
o
.
6
,
p
p
.
1
1
5
4
-
1
1
6
0
,
2
0
1
2
,
d
o
i:
1
0
.
1
0
1
6
/
j.
d
s
p
.
2
0
1
2
.
0
5
.
0
0
7
.
[2
4
]
Ö.
T
u
rg
u
t,
“
A
No
v
e
l
F
e
a
tu
re
S
e
lec
ti
o
n
M
e
th
o
d
f
o
r
S
p
e
e
c
h
Em
o
t
io
n
Re
c
o
g
n
it
io
n
,”
Ap
p
li
e
d
Aco
u
s
ti
c
s
,
v
o
l.
1
4
6
,
p
p
.
3
2
0
-
3
2
6
,
2
0
1
9
,
d
o
i:
1
0
.
1
0
1
6
/
j
.
a
p
a
c
o
u
st.
2
0
1
8
.
1
1
.
0
2
8
.
[2
5
]
S
.
K.
Da
v
is,
M
.
M
o
rn
in
g
sta
r,
M
.
A
.
Dirk
s,
a
n
d
P
.
Qu
a
l
ter
,
“
A
b
il
it
y
E
m
o
ti
o
n
a
l
In
telli
g
e
n
c
e
:
W
h
a
t
a
b
o
u
t
Re
c
o
g
n
it
io
n
o
f
Em
o
ti
o
n
in
Vo
ic
e
s?
,
”
Per
so
n
a
li
ty
a
n
d
In
d
ivid
u
a
l
Diff
e
re
n
c
e
s
,
v
o
l.
1
6
0
,
2
0
2
0
,
A
rt.
n
o
.
1
0
9
9
3
8
,
d
o
i:
1
0
.
1
0
1
6
/j
.
p
a
id
.
2
0
2
0
.
1
0
9
9
3
8
.
[2
6
]
C.
S
.
L
u
ís
a
n
d
C.
F
.
L
im
a
,
“
Re
c
o
g
n
izin
g
Em
o
ti
o
n
s
i
n
S
p
o
k
e
n
L
a
n
g
u
a
g
e
:
A
V
a
li
d
a
ted
S
e
t
o
f
P
o
rt
u
g
u
e
se
S
e
n
ten
c
e
s
a
n
d
P
se
u
d
o
-
se
n
ten
c
e
s
f
o
r
R
e
se
a
r
c
h
o
n
Em
o
ti
o
n
a
l
P
ro
so
d
y
,”
Beh
a
v
io
r
Res
e
a
rc
h
M
e
th
o
d
s
,
v
o
l.
4
2
,
n
o
.
1
,
p
p
.
7
4
-
8
1
,
2
0
1
0
,
d
o
i:
1
0
.
3
7
5
8
/BR
M
.
4
2
.
1
.
7
4
.
[2
7
]
K.
H
.
H
y
u
n
,
E.
H.
Ki
m
,
a
n
d
Y.
K
K
w
a
k
,
“
E
m
o
ti
o
n
a
l
F
e
a
tu
re
Ex
trac
ti
o
n
M
e
th
o
d
Ba
se
d
o
n
th
e
Co
n
c
e
n
trati
o
n
o
f
P
h
o
n
e
m
e
In
f
lu
e
n
c
e
f
o
r
Hu
m
a
n
-
Ro
b
o
t
In
tera
c
ti
o
n
,”
Ad
v
a
n
c
e
d
Ro
b
o
ti
c
s
,
v
o
l.
2
4
,
n
o
.
1
-
2
,
p
p
.
4
7
-
6
7
,
2
0
1
0
,
d
o
i:
1
0
.
1
1
6
3
/0
1
6
9
1
8
6
0
9
X1
2
5
8
5
5
3
0
4
8
7
8
2
2
.
[2
8
]
D
.
Ka
m
in
sk
a
,
T
.
S
a
p
in
sk
i,
a
n
d
G
.
A
n
b
a
rja
f
a
ri
,
“
E
ff
icie
n
c
y
o
f
C
h
o
se
n
S
p
e
e
c
h
De
sc
rip
to
rs
in
Re
l
a
ti
o
n
t
o
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
,”
EURA
S
I
P
J
o
u
r
n
a
l
o
n
A
u
d
i
o
,
S
p
e
e
c
h
,
a
n
d
M
u
sic
Pro
c
e
ss
in
g
,
v
o
l.
2
0
1
7
,
n
o
.
1
,
2
0
1
7
,
d
o
i:
1
0
.
1
1
8
6
/s1
3
6
3
6
-
0
1
7
-
0
1
0
0
-
x.
[2
9
]
A
.
M
.
Be
r
k
e
h
a
n
a
n
d
K.
O
ğ
u
z
,
“
S
p
e
e
c
h
E
m
o
ti
o
n
Re
c
o
g
n
it
i
o
n
:
Em
o
ti
o
n
a
l
M
o
d
e
ls,
Da
tab
a
se
s,
F
e
a
tu
re
s,
P
re
p
ro
c
e
ss
in
g
M
e
th
o
d
s,
S
u
p
p
o
rt
in
g
M
o
d
a
li
ti
e
s,
a
n
d
Clas
sif
iers
,”
S
p
e
e
c
h
Co
mm
u
n
ica
ti
o
n
,
v
o
l.
1
1
6
,
p
p
.
5
6
-
7
6
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
1
6
/
j.
sp
e
c
o
m
.
2
0
1
9
.
1
2
.
0
0
1
.
[3
0
]
M.
E
l
Ay
a
d
i,
M
.
S
.
Ka
m
e
l,
a
n
d
F
.
Ka
rra
y
,
“
S
u
rv
e
y
o
n
S
p
e
e
c
h
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
:
F
e
a
tu
re
s,
Clas
si
f
ica
ti
o
n
S
c
h
e
m
e
s,
a
n
d
Da
tab
a
se
s,”
Pa
tt
e
rn
Rec
o
g
n
it
i
o
n
,
v
o
l.
4
4
,
n
o
.
3
,
p
p
.
572
-
5
8
7
,
2
0
1
1
,
d
o
i:
1
0
.
1
0
1
6
/j
.
p
a
tco
g
.
2
0
1
0
.
0
9
.
0
2
0
.
[3
1
]
Y.
D.
Ch
a
v
h
a
n
,
B.
S
.
Ye
lu
re
,
a
n
d
K.
N.
T
a
y
a
d
e
,
“
S
p
e
e
c
h
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
Us
in
g
RBF
Ke
rn
e
l
o
f
L
IBS
V
M
,
”
2
0
1
5
2
n
d
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
El
e
c
tro
n
ics
a
n
d
Co
mm
u
n
ica
ti
o
n
S
y
ste
ms
(
ICECS
),
2015
,
p
p
.
1
1
3
2
-
11
35
,
d
o
i:
1
0
.
1
1
0
9
/E
CS
.
2
0
1
5
.
7
1
2
4
7
6
0
.
[3
2
]
F
.
A
lb
u
,
D.
Ha
g
ies
c
u
,
L
.
V
lad
u
tu
,
a
n
d
M
.
A
.
P
u
ica
,
“
N
e
u
ra
l
Ne
t
w
o
rk
a
p
p
ro
a
c
h
e
s
f
o
r
c
h
il
d
re
n
’s
e
m
o
ti
o
n
re
c
o
g
n
it
io
n
i
n
i
n
telli
g
e
n
t
lea
rn
in
g
a
p
p
li
c
a
ti
o
n
s,
”
7
th
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Ed
u
c
a
ti
o
n
a
n
d
Ne
w
L
e
a
rn
i
n
g
T
e
c
h
n
o
l
o
g
ies
,
2
0
1
5
,
p
p
.
3
2
2
9
-
3
2
3
9
.
[3
3
]
T
.
L
.
N
w
e
,
S
.
W
.
F
o
o
,
a
n
d
L
.
C.
De
S
il
v
a
,
“
S
p
e
e
c
h
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
Us
in
g
Hi
d
d
e
n
M
a
rk
o
v
M
o
d
e
ls,
”
S
p
e
e
c
h
Co
mm
u
n
ica
ti
o
n
,
v
o
l
.
4
1
,
n
o
.
4
,
p
p
.
6
0
3
-
6
2
3
,
2
0
0
3
,
d
o
i:
1
0
.
1
0
1
6
/S
0
1
6
7
-
6
3
9
3
(
0
3
)
0
0
0
9
9
-
2.
[3
4
]
Z
.
W
a
n
g
,
R.
Jia
o
,
a
n
d
H.
Jia
n
g
,
“
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
Us
in
g
WT
-
S
V
M
i
n
Hu
m
a
n
-
Co
m
p
u
ter
In
tera
c
ti
o
n
,
”
J
o
u
r
n
a
l
o
f
Ne
w
M
e
d
ia
,
v
o
l.
2
,
n
o
.
3
,
p
p
.
1
2
1
-
1
3
0
,
2
0
2
0
,
d
o
i:
1
0
.
3
2
6
0
4
/j
n
m
.
2
0
2
0
.
0
1
0
6
7
4
.
[3
5
]
W
.
Zh
a
n
g
,
X
.
M
e
n
g
,
Z.
L
i,
Q.
L
u
,
a
n
d
S
.
T
a
n
,
“
Em
o
ti
o
n
Re
c
o
g
n
i
ti
o
n
i
n
S
p
e
e
c
h
Us
in
g
M
u
lt
i
-
Clas
sif
ica
ti
o
n
S
V
M
,
”
2
0
1
5
IEE
E
1
2
th
In
tl
Co
n
f
o
n
Ub
i
q
u
it
o
u
s
I
n
telli
g
e
n
c
e
a
n
d
Co
m
p
u
t
i
n
g
a
n
d
2
0
1
5
IEE
E
1
2
th
I
n
tl
C
o
n
f
o
n
A
u
t
o
n
o
mic
a
n
d
T
ru
ste
d
Co
mp
u
ti
n
g
a
n
d
2
0
1
5
IE
EE
1
5
t
h
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
S
c
a
la
b
le
Co
mp
u
t
in
g
a
n
d
Co
mm
u
n
ica
ti
o
n
s
a
n
d
Its
Asso
c
i
a
ted
W
o
rk
sh
o
p
s
(
UIC
-
AT
C
-
S
c
a
l
Co
m)
,
2
0
1
5
,
p
p
.
1
1
8
1
-
1
1
8
6
,
d
o
i
:
1
0
.
1
1
0
9
/UIC
-
AT
C
-
S
c
a
lCo
m
-
CBDC
o
m
-
Io
P
.
2
0
1
5
.
2
1
5
.
[3
6
]
D.
Ba
laji,
“
S
p
e
e
c
h
Em
o
ti
o
n
Re
c
o
g
n
it
io
n
Us
i
n
g
De
e
p
Ne
u
ra
l
Ne
tw
o
rk
s
,”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
f
o
r
Res
e
a
rc
h
i
n
Ap
p
li
e
d
S
c
ien
c
e
a
n
d
En
g
in
e
e
rin
g
T
e
c
h
n
o
lo
g
y
(
IJ
RA
S
ET
)
,
v
o
l.
8
,
n
o
.
6
,
p
p
.
2
4
6
0
-
2
4
6
5
,
2
0
2
0
,
d
o
i:
1
0
.
2
2
2
1
4
/
ij
ra
se
t.
2
0
2
0
.
5
3
5
9
.
[3
7
]
L
.
Ch
e
n
,
W
.
S
u
,
Y.
F
e
n
g
,
M
.
W
u
,
J.
S
h
e
,
a
n
d
K.
Hir
o
ta,
“
T
w
o
-
La
y
e
r
F
u
z
z
y
M
u
lt
ip
le
Ra
n
d
o
m
F
o
re
st
f
o
r
S
p
e
e
c
h
Em
o
ti
o
n
Re
c
o
g
n
it
i
o
n
i
n
Hu
m
a
n
-
Ro
b
o
t
In
tera
c
ti
o
n
,
”
I
n
fo
rm
a
ti
o
n
S
c
ien
c
e
s
,
v
o
l.
5
0
9
,
p
p
.
1
5
0
-
1
6
3
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
1
6
/j
.
in
s.
2
0
1
9
.
0
9
.
0
0
5
.
[3
8
]
H.
Ka
y
a
a
n
d
A
.
A
.
Ka
rp
o
v
,
“
E
ff
icie
n
t
a
n
d
e
ff
e
c
ti
v
e
stra
teg
i
e
s
f
o
r
c
ro
ss
-
c
o
rp
u
s
a
c
o
u
stic
e
m
o
ti
o
n
re
c
o
g
n
it
io
n
,
”
Ne
u
ro
c
o
mp
u
ti
n
g
,
v
o
l
.
2
7
5
,
p
p
.
1
0
2
8
-
1
0
3
4
,
2
0
1
8
,
d
o
i:
1
0
.
1
0
1
6
/j
.
n
e
u
c
o
m
.
2
0
1
7
.
0
9
.
0
4
9
.
[3
9
]
K.
Ha
n
,
D.
Yu
,
a
n
d
I.
T
a
sh
e
v
,
“
S
p
e
e
c
h
e
m
o
ti
o
n
re
c
o
g
n
it
io
n
u
s
in
g
d
e
e
p
n
e
u
ra
l
n
e
tw
o
rk
a
n
d
e
x
trem
e
lea
rn
in
g
m
a
c
h
in
e
,
”
Fi
ft
e
e
n
th
a
n
n
u
a
l
c
o
n
fe
re
n
c
e
o
f
th
e
i
n
ter
n
a
ti
o
n
a
l
sp
e
e
c
h
c
o
mm
u
n
ica
t
io
n
a
ss
o
c
i
a
ti
o
n
,
2
0
1
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.