<s>
Imagined	B-Algorithm
speech	I-Algorithm
(	O
also	O
called	O
silent	B-Algorithm
speech	I-Algorithm
,	O
covert	O
speech	O
,	O
inner	O
speech	O
,	O
or	O
,	O
in	O
the	O
original	O
Latin	O
terminology	O
used	O
by	O
clinicians	O
,	O
endophasia	B-Algorithm
)	O
is	O
thinking	O
in	O
the	O
form	O
of	O
sound	O
–	O
“	O
hearing	O
”	O
one	O
’s	O
own	O
voice	O
silently	O
to	O
oneself	O
,	O
without	O
the	O
intentional	O
movement	O
of	O
any	O
extremities	O
such	O
as	O
the	O
lips	O
,	O
tongue	O
,	O
or	O
hands	O
.	O
</s>
<s>
Logically	O
,	O
imagined	B-Algorithm
speech	I-Algorithm
has	O
been	O
possible	O
since	O
the	O
emergence	O
of	O
language	O
,	O
however	O
,	O
the	O
phenomenon	O
is	O
most	O
associated	O
with	O
its	O
investigation	O
through	O
signal	O
processing	O
and	O
detection	O
within	O
electroencephalograph	B-Application
(	O
EEG	B-Application
)	O
data	O
as	O
well	O
as	O
data	O
obtained	O
using	O
alternative	O
non-invasive	O
,	O
brain	B-Application
–	I-Application
computer	I-Application
interface	I-Application
(	O
BCI	O
)	O
devices	O
.	O
</s>
<s>
These	O
imagined	B-Algorithm
speech	I-Algorithm
signals	O
would	O
be	O
analyzed	O
and	O
translated	O
into	O
distinct	O
words	O
allowing	O
covert	O
person-to-person	O
communication.	O
”	O
In	O
his	O
"	O
Impossible	O
languages	O
"	O
(	O
2016	O
)	O
Andrea	O
Moro	O
discusses	O
the	O
"	O
sound	O
of	O
thoughts	O
"	O
and	O
the	O
relationship	O
between	O
linguistics	O
units	O
and	O
imagined	B-Algorithm
speech	I-Algorithm
,	O
mainly	O
capitalizing	O
on	O
Magrassi	O
et	O
al	O
.	O
</s>
<s>
The	O
process	O
for	O
analyzing	O
subjects	O
 '	O
silent	B-Algorithm
speech	I-Algorithm
is	O
composed	O
of	O
recording	O
subjects’	O
brain	B-Algorithm
waves	I-Algorithm
,	O
and	O
then	O
using	O
a	O
computer	O
to	O
process	O
the	O
data	O
and	O
determine	O
the	O
content	O
of	O
the	O
subjects	O
 '	O
covert	O
speech	O
.	O
</s>
<s>
Subject	O
neural	O
patterns	O
(	O
brain	B-Algorithm
waves	I-Algorithm
)	O
can	O
be	O
recorded	O
using	O
BCI	O
devices	O
;	O
currently	O
,	O
use	O
of	O
non-invasive	O
devices	O
,	O
specifically	O
the	O
EEG	B-Application
,	O
is	O
of	O
greater	O
interest	O
to	O
researchers	O
than	O
invasive	O
and	O
partially	O
invasive	O
types	O
.	O
</s>
<s>
This	O
is	O
because	O
non-invasive	O
types	O
pose	O
the	O
least	O
risk	O
to	O
subject	O
health	O
;	O
EEG	B-Application
's	O
have	O
attracted	O
the	O
greatest	O
interest	O
because	O
they	O
offer	O
the	O
most	O
user-friendly	O
approach	O
in	O
addition	O
to	O
having	O
far	O
less	O
complex	O
instrumentation	B-Application
than	O
that	O
of	O
functional	B-Algorithm
magnetic	I-Algorithm
resonance	I-Algorithm
imaging	I-Algorithm
(	O
fMRI	B-Algorithm
’s	O
)	O
,	O
another	O
commonly	O
used	O
non-invasive	O
BCI	O
.	O
</s>
<s>
After	O
artifact-removal	O
,	O
a	O
series	O
of	O
algorithms	O
is	O
used	O
to	O
translate	O
raw	O
data	O
into	O
the	O
imagined	B-Algorithm
speech	I-Algorithm
content	O
.	O
</s>
<s>
Presumably	O
,	O
“	O
thinking	O
in	O
the	O
form	O
of	O
sound	O
”	O
recruits	O
auditory	O
and	O
language	O
areas	O
whose	O
activation	O
profiles	O
may	O
be	O
extracted	O
from	O
the	O
EEG	B-Application
,	O
given	O
adequate	O
processing	O
.	O
</s>
<s>
Such	O
linear	O
mapping	O
from	O
EEG	B-Application
to	O
stimulus	O
is	O
an	O
example	O
of	O
neural	O
decoding	O
.	O
</s>
<s>
Hence	O
one	O
can	O
have	O
the	O
same	O
EEG	B-Application
signal	O
,	O
but	O
it	O
is	O
uncertain	O
,	O
at	O
least	O
in	O
acoustic	O
terms	O
,	O
what	O
stimulus	O
to	O
map	O
it	O
to	O
.	O
</s>
<s>
In	O
detection	O
of	O
other	O
imagined	O
actions	O
,	O
such	O
as	O
imagined	O
physical	O
movements	O
,	O
greater	O
brain	B-Application
activity	I-Application
occurs	O
in	O
one	O
hemisphere	O
over	O
the	O
other	O
.	O
</s>
<s>
In	O
imagined	B-Algorithm
speech	I-Algorithm
detection	O
,	O
equal	O
levels	O
of	O
activity	O
commonly	O
occur	O
in	O
both	O
the	O
left	O
and	O
right	O
hemispheres	O
simultaneously	O
.	O
</s>
<s>
Artifacts	O
present	O
in	O
EEG	B-Application
data	O
are	O
just	O
one	O
of	O
many	O
significant	O
sources	O
of	O
noise	O
.	O
</s>
<s>
To	O
further	O
complicate	O
matters	O
,	O
the	O
relative	O
placement	O
of	O
EEG	B-Application
electrodes	O
will	O
vary	O
amongst	O
subjects	O
.	O
</s>
<s>
This	O
is	O
because	O
the	O
anatomical	O
details	O
of	O
people	O
's	O
heads	O
will	O
differ	O
;	O
therefore	O
,	O
the	O
signals	O
recorded	O
will	O
vary	O
in	O
each	O
subject	O
,	O
regardless	O
of	O
individuals-specific	O
imagined	B-Algorithm
speech	I-Algorithm
characteristics	O
.	O
</s>
