<s>
Signal	O
processing	O
is	O
an	O
electrical	O
engineering	O
subfield	O
that	O
focuses	O
on	O
analysing	O
,	O
modifying	O
,	O
and	O
synthesizing	O
signals	O
such	O
as	O
sound	B-Algorithm
,	O
images	B-Algorithm
,	O
and	O
scientific	O
measurements	O
.	O
</s>
<s>
For	O
example	O
,	O
with	O
a	O
filter	O
g	O
,	O
an	O
inverse	B-Application
filter	I-Application
h	O
is	O
one	O
such	O
that	O
the	O
sequence	O
of	O
applying	O
g	O
then	O
h	O
to	O
a	O
signal	O
results	O
in	O
the	O
original	O
signal	O
.	O
</s>
<s>
Software	O
or	O
electronic	O
inverse	B-Application
filters	I-Application
are	O
often	O
used	O
to	O
compensate	O
for	O
the	O
effect	O
of	O
unwanted	O
environmental	O
filtering	O
of	O
signals	O
.	O
</s>
<s>
In	O
all	O
proposed	O
models	O
for	O
the	O
production	O
of	O
human	O
speech	O
,	O
an	O
important	O
variable	O
is	O
the	O
waveform	O
of	O
the	O
airflow	O
,	O
or	O
volume	O
velocity	O
,	O
at	O
the	O
glottis	B-Application
.	O
</s>
<s>
The	O
glottal	O
volume	O
velocity	O
waveform	O
provides	O
the	O
link	O
between	O
movements	O
of	O
the	O
vocal	O
folds	O
and	O
the	O
acoustical	O
results	O
of	O
such	O
movements	O
,	O
in	O
that	O
the	O
glottis	B-Application
acts	O
approximately	O
as	O
a	O
source	O
of	O
volume	O
velocity	O
.	O
</s>
<s>
That	O
is	O
,	O
the	O
impedance	O
of	O
the	O
glottis	B-Application
is	O
usually	O
much	O
higher	O
than	O
that	O
of	O
the	O
vocal	O
tract	O
,	O
and	O
so	O
glottal	O
airflow	O
is	O
controlled	O
mostly	O
(	O
but	O
not	O
entirely	O
)	O
by	O
glottal	O
area	O
and	O
subglottal	O
pressure	O
,	O
and	O
not	O
by	O
vocal-tract	O
acoustics	O
.	O
</s>
<s>
This	O
view	O
of	O
voiced	O
speech	O
production	O
is	O
often	O
referred	O
to	O
as	O
the	O
source-filter	B-Application
model	I-Application
.	O
</s>
<s>
A	O
technique	O
for	O
obtaining	O
an	O
estimate	O
of	O
the	O
glottal	O
volume	O
velocity	O
waveform	O
during	O
voiced	O
speech	O
is	O
the	O
“	O
inverse-filtering	O
”	O
of	O
either	O
the	O
radiated	O
acoustic	O
waveform	O
,	O
as	O
measured	O
by	O
a	O
microphone	B-Application
having	O
a	O
good	O
low	O
frequency	O
response	O
,	O
or	O
the	O
volume	O
velocity	O
at	O
the	O
mouth	O
,	O
as	O
measured	O
by	O
a	O
pneumotachograph	O
at	O
the	O
mouth	O
having	O
a	O
linear	O
response	O
,	O
little	O
speech	O
distortion	O
,	O
and	O
a	O
response	O
time	O
of	O
under	O
approximately	O
1/2	O
ms.	O
A	O
pneumotachograph	O
having	O
these	O
properties	O
was	O
first	O
described	O
by	O
Rothenberg	O
and	O
termed	O
by	O
him	O
a	O
circumferentially	O
vented	O
mask	O
or	O
CV	O
mask	O
.	O
</s>
<s>
The	O
transfer	O
characteristic	O
of	O
the	O
supraglottal	O
vocal	O
tract	O
is	O
defined	O
with	O
the	O
input	O
to	O
the	O
vocal	O
tract	O
considered	O
to	O
be	O
the	O
volume	O
velocity	O
at	O
the	O
glottis	B-Application
.	O
</s>
<s>
For	O
non-nasalized	O
vowels	O
,	O
assuming	O
a	O
high-impedance	O
volume	O
velocity	O
source	O
at	O
the	O
glottis	B-Application
,	O
the	O
transfer	O
function	O
of	O
the	O
vocal	O
tract	O
below	O
about	O
3000Hz	O
contains	O
a	O
number	O
of	O
pairs	O
of	O
complex-conjugate	O
poles	O
,	O
more	O
commonly	O
referred	O
to	O
as	O
resonances	O
or	O
formants	O
.	O
</s>
<s>
If	O
the	O
input	O
is	O
from	O
a	O
microphone	B-Application
,	O
and	O
not	O
a	O
CV	O
mask	O
or	O
its	O
equivalent	O
,	O
the	O
inverse	B-Application
filter	I-Application
also	O
must	O
have	O
a	O
pole	O
at	O
zero	O
frequency	O
(	O
an	O
integration	O
operation	O
)	O
to	O
account	O
for	O
the	O
radiation	O
characteristic	O
that	O
connects	O
volume	O
velocity	O
with	O
acoustic	O
pressure	O
.	O
</s>
<s>
Inverse	O
filtering	O
the	O
output	O
of	O
a	O
CV	O
mask	O
retains	O
the	O
level	O
of	O
zero	O
flow	O
,	O
while	O
inverse	O
filtering	O
a	O
microphone	B-Application
signal	O
does	O
not	O
.	O
</s>
<s>
Inverse	O
filtering	O
depends	O
on	O
the	O
source-filter	B-Application
model	I-Application
and	O
a	O
vocal	O
tract	O
filter	O
that	O
is	O
linear	O
system	O
,	O
however	O
,	O
the	O
source	O
and	O
filter	O
need	O
not	O
be	O
independent	O
.	O
</s>
