<s>
Temporal	O
envelope	B-Algorithm
(	O
ENV	O
)	O
and	O
temporal	B-Application
fine	I-Application
structure	I-Application
(	O
TFS	O
)	O
are	O
changes	O
in	O
the	O
amplitude	B-Application
and	O
frequency	O
of	O
sound	O
perceived	O
by	O
humans	O
over	O
time	O
.	O
</s>
<s>
These	O
temporal	O
changes	O
are	O
responsible	O
for	O
several	O
aspects	O
of	O
auditory	O
perception	O
,	O
including	O
loudness	B-Algorithm
,	O
pitch	O
and	O
timbre	O
perception	O
and	O
spatial	O
hearing	O
.	O
</s>
<s>
A	O
dichotomy	O
between	O
slow	O
"	O
temporal	O
envelope	B-Algorithm
"	O
cues	O
and	O
faster	O
"	O
temporal	B-Application
fine	I-Application
structure	I-Application
"	O
cues	O
has	O
been	O
proposed	O
to	O
study	O
several	O
aspects	O
of	O
auditory	O
perception	O
(	O
e.g.	O
,	O
loudness	B-Algorithm
,	O
pitch	O
and	O
timbre	O
perception	O
,	O
auditory	B-Application
scene	I-Application
analysis	I-Application
,	O
sound	O
localization	O
)	O
at	O
two	O
distinct	O
time	O
scales	O
in	O
each	O
frequency	O
band	O
.	O
</s>
<s>
Over	O
the	O
last	O
decades	O
,	O
a	O
wealth	O
of	O
psychophysical	O
,	O
electrophysiological	O
and	O
computational	O
studies	O
based	O
on	O
this	O
envelope/fine	O
-structure	O
dichotomy	O
have	O
examined	O
the	O
role	O
of	O
these	O
temporal	O
cues	O
in	O
sound	O
identification	O
and	O
communication	O
,	O
how	O
these	O
temporal	O
cues	O
are	O
processed	O
by	O
the	O
peripheral	O
and	O
central	O
auditory	O
system	O
,	O
and	O
the	O
effects	O
of	O
aging	O
and	O
cochlear	O
damage	O
on	O
temporal	O
auditory	O
processing	O
.	O
</s>
<s>
Although	O
the	O
envelope/fine	O
-structure	O
dichotomy	O
has	O
been	O
debated	O
and	O
questions	O
remain	O
as	O
to	O
how	O
temporal	B-Application
fine	I-Application
structure	I-Application
cues	O
are	O
actually	O
encoded	O
in	O
the	O
auditory	O
system	O
,	O
these	O
studies	O
have	O
led	O
to	O
a	O
range	O
of	O
applications	O
in	O
various	O
fields	O
including	O
speech	O
and	O
audio	O
processing	O
,	O
clinical	O
audiology	O
and	O
rehabilitation	O
of	O
sensorineural	O
hearing	O
loss	O
via	O
hearing	O
aids	O
or	O
cochlear	O
implants	O
.	O
</s>
<s>
Notions	O
of	O
temporal	B-Application
envelope	I-Application
and	I-Application
temporal	I-Application
fine	I-Application
structure	I-Application
may	O
have	O
different	O
meanings	O
in	O
many	O
studies	O
.	O
</s>
<s>
Any	O
sound	O
whose	O
frequency	O
components	O
cover	O
a	O
narrow	O
range	O
(	O
called	O
a	O
narrowband	O
signal	O
)	O
can	O
be	O
considered	O
as	O
an	O
envelope	B-Algorithm
(	O
ENVp	O
,	O
where	O
p	O
denotes	O
the	O
physical	O
signal	O
)	O
superimposed	O
on	O
a	O
more	O
rapidly	O
oscillating	O
carrier	O
,	O
the	O
temporal	B-Application
fine	I-Application
structure	I-Application
(	O
TFSp	O
)	O
.	O
</s>
<s>
Therefore	O
,	O
the	O
waveform	O
at	O
each	O
place	O
on	O
the	O
BM	O
can	O
be	O
considered	O
as	O
an	O
envelope	B-Algorithm
(	O
ENVBM	O
)	O
superimposed	O
on	O
a	O
more	O
rapidly	O
oscillating	O
carrier	O
,	O
the	O
temporal	B-Application
fine	I-Application
structure	I-Application
(	O
TFSBM	O
)	O
.	O
</s>
<s>
Both	O
ENVBM	O
and	O
TFSBM	O
are	O
represented	O
in	O
the	O
time	O
patterns	O
of	O
action	B-Algorithm
potentials	I-Algorithm
in	O
the	O
auditory	O
nerve	O
these	O
are	O
denoted	O
ENVn	O
and	O
TFSn	O
.	O
</s>
<s>
The	O
neural	O
representation	O
of	O
stimulus	O
envelope	B-Algorithm
,	O
ENVn	O
,	O
has	O
typically	O
been	O
studied	O
using	O
well-controlled	O
ENVp	O
modulations	O
,	O
that	O
is	O
sinusoidally	O
amplitude-modulated	B-Device
(	O
AM	O
)	O
sounds	O
.	O
</s>
<s>
Some	O
of	O
these	O
cells	O
show	O
an	O
excellent	O
response	O
to	O
the	O
ENVn	O
and	O
provide	O
inhibitory	O
sideband	O
inputs	O
to	O
other	O
cells	O
in	O
the	O
cochlear	O
nucleus	O
giving	O
a	O
physiological	O
correlate	O
of	O
comodulation	O
masking	O
release	O
,	O
a	O
phenomenon	O
whereby	O
the	O
detection	O
of	O
a	O
signal	O
in	O
a	O
masker	O
is	O
improved	O
when	O
the	O
masker	O
has	O
correlated	O
envelope	B-Algorithm
fluctuations	O
across	O
frequency	O
(	O
see	O
section	O
below	O
)	O
.	O
</s>
<s>
Responses	O
to	O
the	O
temporal-envelope	O
cues	O
of	O
speech	O
or	O
other	O
complex	O
sounds	O
persist	O
up	O
the	O
auditory	O
pathway	O
,	O
eventually	O
to	O
the	O
various	O
fields	O
of	O
the	O
auditory	O
cortex	O
in	O
many	O
animals	O
.	O
</s>
<s>
Tuning	O
to	O
AM	O
rates	O
below	O
about	O
64Hz	O
is	O
also	O
found	O
in	O
the	O
human	O
auditory	O
cortex	O
as	O
revealed	O
by	O
brain-imaging	O
techniques	O
(	O
fMRI	B-Algorithm
)	O
and	O
cortical	O
recordings	O
in	O
epileptic	O
patients	O
(	O
electrocorticography	O
)	O
.	O
</s>
<s>
The	O
ranges	O
over	O
which	O
cortical	O
responses	O
encode	O
well	O
the	O
temporal-envelope	O
cues	O
of	O
speech	O
have	O
been	O
shown	O
to	O
be	O
predictive	O
of	O
the	O
human	O
ability	O
to	O
understand	O
speech	O
.	O
</s>
<s>
One	O
unexpected	O
aspect	O
of	O
phase	O
locking	O
in	O
the	O
auditory	O
cortex	O
has	O
been	O
observed	O
in	O
the	O
responses	O
elicited	O
by	O
complex	O
acoustic	O
stimuli	O
with	O
spectrograms	O
that	O
exhibit	O
relatively	O
slow	O
envelopes	B-Algorithm
(	O
<	O
20Hz	O
)	O
,	O
but	O
that	O
are	O
carried	O
by	O
fast	O
modulations	O
that	O
are	O
as	O
high	O
as	O
hundreds	O
of	O
Hertz	O
.	O
</s>
<s>
For	O
these	O
stimuli	O
,	O
cortical	O
responses	O
phase-lock	O
to	O
both	O
the	O
envelope	B-Algorithm
and	O
fine-structure	O
induced	O
by	O
interactions	O
between	O
unresolved	O
harmonics	O
of	O
the	O
sound	O
,	O
thus	O
reflecting	O
the	O
pitch	O
of	O
the	O
sound	O
,	O
and	O
exceeding	O
the	O
typical	O
lower	O
limits	O
of	O
cortical	O
phase-locking	O
to	O
the	O
envelopes	B-Algorithm
of	O
a	O
few	O
10	O
's	O
of	O
Hertz	O
.	O
</s>
<s>
Low	O
rates	O
of	O
AM	O
,	O
in	O
the	O
1	O
–	O
8Hz	O
range	O
,	O
are	O
perceived	O
as	O
changes	O
in	O
perceived	O
intensity	O
,	O
that	O
is	O
loudness	B-Algorithm
fluctuations	O
(	O
a	O
percept	O
that	O
can	O
also	O
be	O
evoked	O
by	O
frequency	B-Application
modulation	I-Application
,	O
FM	B-Application
)	O
;	O
at	O
higher	O
rates	O
,	O
AM	O
is	O
perceived	O
as	O
roughness	O
,	O
with	O
the	O
greatest	O
roughness	O
sensation	O
occurring	O
at	O
around	O
70Hz	O
;	O
at	O
even	O
higher	O
rates	O
,	O
AM	O
can	O
evoke	O
a	O
weak	O
pitch	O
percept	O
corresponding	O
to	O
the	O
modulation	O
rate	O
.	O
</s>
<s>
Correlated	O
envelope	B-Algorithm
fluctuations	O
across	O
frequency	O
in	O
a	O
masker	O
can	O
aid	O
detection	O
of	O
a	O
pure	O
tone	O
signal	O
,	O
an	O
effect	O
known	O
as	O
comodulation	O
masking	O
release	O
.	O
</s>
<s>
These	O
cues	O
arise	O
from	O
the	O
interaction	O
of	O
different	O
modulation	O
rates	O
,	O
previously	O
described	O
as	O
"	O
beating	O
"	O
in	O
the	O
envelope-frequency	O
domain	O
.	O
</s>
<s>
Perception	O
of	O
second-order	O
AM	O
has	O
been	O
interpreted	O
as	O
resulting	O
from	O
nonlinear	O
mechanisms	O
in	O
the	O
auditory	O
pathway	O
that	O
produce	O
an	O
audible	O
distortion	O
component	O
at	O
the	O
envelope	B-Algorithm
beat	O
frequency	O
in	O
the	O
internal	O
modulation	O
spectrum	O
of	O
the	O
sounds	O
.	O
</s>
<s>
Interaural	O
time	O
differences	O
in	O
the	O
envelope	B-Algorithm
provide	O
binaural	O
cues	O
even	O
at	O
high	O
frequencies	O
where	O
TFSn	O
cannot	O
be	O
used	O
.	O
</s>
<s>
This	O
model	O
extracts	B-Algorithm
the	I-Algorithm
temporal	I-Algorithm
envelope	I-Algorithm
of	O
the	O
sound	O
(	O
ENVp	O
)	O
via	O
bandpass	O
filtering	O
,	O
half-wave	O
rectification	O
(	O
which	O
may	O
be	O
followed	O
by	O
fast-acting	O
amplitude	B-Application
compression	O
)	O
,	O
and	O
lowpass	O
filtering	O
with	O
a	O
cutoff	O
frequency	O
between	O
about	O
60	O
and	O
150Hz	O
.	O
</s>
<s>
The	O
leaky	O
integrator	O
is	O
often	O
used	O
with	O
a	O
decision	O
statistic	O
based	O
on	O
either	O
the	O
resulting	O
envelope	B-Algorithm
power	O
,	O
the	O
max/min	O
ratio	O
,	O
or	O
the	O
crest	O
factor	O
.	O
</s>
<s>
Based	O
on	O
the	O
concept	O
of	O
frequency	O
selectivity	O
for	O
AM	O
,	O
the	O
perception	O
model	O
of	O
Torsten	O
Dau	O
incorporates	O
broadly	O
tuned	O
bandpass	O
modulation	O
filters	O
(	O
with	O
a	O
Q	O
value	O
around	O
1	O
)	O
to	O
account	O
for	O
data	O
from	O
a	O
broad	O
variety	O
of	O
psychoacoustic	O
tasks	O
and	O
particularly	O
AM	O
detection	O
for	O
noise	O
carriers	O
with	O
different	O
bandwidths	O
,	O
taking	O
into	O
account	O
their	O
intrinsic	O
envelope	B-Algorithm
fluctuations	O
.	O
</s>
<s>
The	O
shapes	O
of	O
the	O
modulation	O
filters	O
have	O
been	O
estimated	O
and	O
an	O
“	O
envelope	B-Algorithm
power	O
spectrum	O
model	O
”	O
(	O
EPSM	O
)	O
based	O
on	O
these	O
filters	O
can	O
account	O
for	O
AM	O
masking	O
patterns	O
and	O
AM	O
depth	O
discrimination	O
.	O
</s>
<s>
The	O
neural	O
representation	O
of	O
temporal	B-Application
fine	I-Application
structure	I-Application
,	O
TFSn	O
,	O
has	O
been	O
studied	O
using	O
stimuli	O
with	O
well-controlled	O
TFSp	O
:	O
pure	O
tones	O
,	O
harmonic	O
complex	O
tones	O
,	O
and	O
frequency-modulated	B-Application
(	O
FM	B-Application
)	O
tones	O
.	O
</s>
<s>
TFSn	O
information	O
in	O
the	O
auditory	O
nerve	O
may	O
be	O
used	O
to	O
encode	O
the	O
(	O
audio	O
)	O
frequency	O
of	O
low-frequency	O
sounds	O
,	O
including	O
single	O
tones	O
and	O
more	O
complex	O
stimuli	O
such	O
as	O
frequency-modulated	B-Application
tones	O
or	O
steady-state	O
vowels	O
(	O
see	O
role	O
and	O
applications	O
to	O
speech	O
and	O
music	O
)	O
.	O
</s>
<s>
These	O
capacities	O
include	O
discrimination	O
of	O
frequency	O
,	O
discrimination	O
of	O
the	O
fundamental	O
frequency	O
of	O
harmonic	O
sounds	O
,	O
detection	O
of	O
FM	B-Application
at	O
rates	O
below	O
5Hz	O
,	O
melody	O
recognition	O
for	O
sequences	O
of	O
pure	O
tones	O
and	O
complex	O
tones	O
,	O
lateralization	O
and	O
localization	O
of	O
pure	O
tones	O
and	O
complex	O
tones	O
,	O
and	O
segregation	O
of	O
concurrent	O
harmonic	O
sounds	O
(	O
such	O
as	O
speech	O
sounds	O
)	O
.	O
</s>
<s>
As	O
for	O
FM	B-Application
detection	O
,	O
the	O
current	O
view	O
assumes	O
that	O
in	O
the	O
normal	O
auditory	O
system	O
,	O
FM	B-Application
is	O
encoded	O
via	O
TFSn	O
cues	O
when	O
the	O
FM	B-Application
rate	O
is	O
low	O
(	O
<	O
5Hz	O
)	O
and	O
when	O
the	O
carrier	O
frequency	O
is	O
below	O
about	O
4kHz	O
,	O
and	O
via	O
ENVn	O
cues	O
when	O
the	O
FM	B-Application
is	O
fast	O
or	O
when	O
the	O
carrier	O
frequency	O
is	O
higher	O
than	O
4kHz	O
.	O
</s>
<s>
According	O
to	O
this	O
view	O
,	O
TFSn	O
cues	O
are	O
not	O
used	O
to	O
detect	O
FM	B-Application
with	O
rates	O
above	O
about	O
10Hz	O
because	O
the	O
mechanism	O
decoding	O
the	O
TFSn	O
information	O
is	O
“	O
sluggish	O
”	O
and	O
cannot	O
track	O
rapid	O
changes	O
in	O
frequency	O
.	O
</s>
<s>
Several	O
studies	O
have	O
shown	O
that	O
auditory	O
sensitivity	O
to	O
slow	O
FM	B-Application
at	O
low	O
carrier	O
frequency	O
is	O
associated	O
with	O
speech	O
identification	O
for	O
both	O
normal-hearing	O
and	O
hearing-impaired	O
individuals	O
when	O
speech	O
reception	O
is	O
limited	O
by	O
acoustic	O
degradations	O
(	O
e.g.	O
,	O
filtering	O
)	O
or	O
concurrent	O
speech	O
sounds	O
.	O
</s>
<s>
The	O
separation	O
of	O
a	O
sound	O
into	O
ENVp	O
and	O
TFSp	O
appears	O
inspired	O
partly	O
by	O
how	O
sounds	O
are	O
synthesized	O
and	O
by	O
the	O
availability	O
of	O
a	O
convenient	O
way	O
to	O
separate	O
an	O
existing	O
sound	O
into	O
ENV	O
and	O
TFS	O
,	O
namely	O
the	O
Hilbert	B-Algorithm
transform	I-Algorithm
.	O
</s>
<s>
There	O
is	O
a	O
risk	O
that	O
this	O
view	O
of	O
auditory	O
processing	O
is	O
dominated	O
by	O
these	O
physical/technical	O
concepts	O
,	O
similarly	O
to	O
how	O
cochlear	O
frequency-to-place	O
mapping	O
was	O
for	O
a	O
long	O
time	O
conceptualized	O
in	O
terms	O
of	O
the	O
Fourier	B-Algorithm
transform	I-Algorithm
.	O
</s>
<s>
These	O
concepts	O
may	O
be	O
related	O
to	O
separating	O
ENV	O
and	O
TFS	O
but	O
are	O
unlikely	O
realized	O
like	O
the	O
Hilbert	B-Algorithm
transform	I-Algorithm
.	O
</s>
<s>
In	O
two	O
simulation	O
studies	O
,	O
the	O
mean-rate	O
and	O
spike-timing	O
information	O
was	O
quantified	O
at	O
the	O
output	O
of	O
such	O
a	O
model	O
to	O
characterize	O
,	O
respectively	O
,	O
the	O
short-term	O
rate	O
of	O
neural	B-Algorithm
firing	I-Algorithm
(	O
ENVn	O
)	O
and	O
the	O
level	O
of	O
synchronization	O
due	O
to	O
phase	O
locking	O
(	O
TFSn	O
)	O
in	O
response	O
to	O
speech	O
sounds	O
degraded	O
by	O
vocoders	B-Application
.	O
</s>
<s>
The	O
best	O
model	O
predictions	O
of	O
vocoded-speech	O
intelligibility	O
were	O
found	O
when	O
both	O
ENVn	O
and	O
TFSn	O
cues	O
were	O
included	O
,	O
providing	O
evidence	O
that	O
TFSn	O
cues	O
are	O
important	O
for	O
intelligibility	O
when	O
the	O
speech	O
ENVp	O
cues	O
are	O
degraded	O
.	O
</s>
<s>
On	O
the	O
other	O
hand	O
,	O
the	O
fundamental	O
frequency	O
(	O
“	O
pitch	O
”	O
)	O
contour	O
of	O
speech	O
sounds	O
is	O
primarily	O
conveyed	O
via	O
TFSp	O
cues	O
,	O
although	O
some	O
information	O
on	O
the	O
contour	O
can	O
be	O
perceived	O
via	O
rapid	O
envelope	B-Algorithm
fluctuations	O
corresponding	O
to	O
the	O
fundamental	O
frequency	O
.	O
</s>
<s>
In	O
addition	O
,	O
several	O
vocoder	B-Application
studies	O
have	O
suggested	O
that	O
TFSp	O
cues	O
contribute	O
to	O
the	O
intelligibility	O
of	O
speech	O
in	O
quiet	O
and	O
noise	O
.	O
</s>
<s>
Early	O
studies	O
highlighted	O
the	O
importance	O
of	O
envelope-based	O
temporal	O
patterning	O
in	O
perception	O
of	O
environmental	O
events	O
.	O
</s>
<s>
For	O
instance	O
,	O
Warren	O
&	O
Verbrugge	O
,	O
demonstrated	O
that	O
constructed	O
sounds	O
of	O
a	O
glass	O
bottle	O
dropped	O
on	O
the	O
floor	O
were	O
perceived	O
as	O
bouncing	O
when	O
high-energy	O
regions	O
in	O
four	O
different	O
frequency	O
bands	O
were	O
temporally	O
aligned	O
,	O
producing	O
amplitude	B-Application
peaks	O
in	O
the	O
envelope	B-Algorithm
.	O
</s>
<s>
More	O
recent	O
studies	O
using	O
vocoder	B-Application
simulations	O
of	O
cochlear	O
implant	O
processing	O
demonstrated	O
that	O
many	O
temporally-patterned	O
sounds	O
can	O
be	O
perceived	O
with	O
little	O
original	O
spectral	O
information	O
,	O
based	O
primarily	O
on	O
temporal	O
cues	O
.	O
</s>
<s>
Such	O
sounds	O
as	O
footsteps	O
,	O
horse	O
galloping	O
,	O
helicopter	O
flying	O
,	O
ping-pong	O
playing	O
,	O
clapping	O
,	O
typing	O
were	O
identified	O
with	O
a	O
high	O
accuracy	O
of	O
70%	O
or	O
more	O
with	O
a	O
single	O
channel	O
of	O
envelope-modulated	O
broadband	O
noise	O
or	O
with	O
only	O
two	O
frequency	O
channels	O
.	O
</s>
<s>
In	O
these	O
studies	O
,	O
envelope-based	O
acoustic	O
measures	O
such	O
as	O
number	O
of	O
bursts	O
and	O
peaks	O
in	O
the	O
envelope	B-Algorithm
were	O
predictive	O
of	O
listeners’	O
abilities	O
to	O
identify	O
sounds	O
based	O
primarily	O
on	O
ENVp	O
cues	O
.	O
</s>
<s>
Listeners	O
with	O
cochlear	O
implants	O
,	O
which	O
transmit	O
envelope	B-Algorithm
information	O
for	O
specific	O
frequency	O
bands	O
,	O
but	O
do	O
not	O
transmit	O
TFSp	O
,	O
have	O
considerably	O
reduced	O
abilities	O
in	O
identification	O
of	O
common	O
environmental	O
sounds	O
.	O
</s>
<s>
When	O
heard	O
within	O
an	O
auditory	O
scene	O
,	O
accurate	O
identification	O
of	O
individual	O
environmental	O
sounds	O
is	O
contingent	O
on	O
the	O
ability	O
to	O
segregate	O
them	O
from	O
other	O
sound	O
sources	O
or	O
auditory	O
streams	O
in	O
the	O
auditory	O
scene	O
,	O
which	O
involves	O
further	O
reliance	O
on	O
ENVp	O
and	O
TFSp	O
cues	O
(	O
see	O
Role	O
in	O
auditory	B-Application
scene	I-Application
analysis	I-Application
)	O
.	O
</s>
<s>
Auditory	B-Application
scene	I-Application
analysis	I-Application
refers	O
to	O
the	O
ability	O
to	O
perceive	O
separately	O
sounds	O
coming	O
from	O
different	O
sources	O
.	O
</s>
<s>
Electrophysiological	O
studies	O
conducted	O
in	O
1-month-old	O
infants	O
using	O
2000Hz	O
AM	O
pure	O
tones	O
indicate	O
some	O
immaturity	O
in	O
envelope	B-Algorithm
following	O
response	O
(	O
EFR	O
)	O
.	O
</s>
<s>
Regarding	O
the	O
development	O
of	O
speech	O
ENVp	O
processing	O
,	O
vocoder	B-Application
studies	O
suggest	O
that	O
infants	O
as	O
young	O
as	O
3	O
months	O
are	O
able	O
to	O
discriminate	O
a	O
change	O
in	O
consonants	O
when	O
the	O
faster	O
ENVp	O
information	O
of	O
the	O
syllables	O
is	O
preserved	O
(	O
<	O
256Hz	O
)	O
but	O
less	O
so	O
when	O
only	O
the	O
slowest	O
ENVp	O
is	O
available	O
(	O
<	O
8Hz	O
)	O
.	O
</s>
<s>
The	O
effects	O
of	O
hearing	O
loss	O
and	O
age	O
on	O
neural	O
coding	O
are	O
generally	O
believed	O
to	O
be	O
smaller	O
for	O
slowly	O
varying	O
envelope	B-Algorithm
responses	O
(	O
i.e.	O
,	O
ENVn	O
)	O
than	O
for	O
rapidly	O
varying	O
temporal	B-Application
fine	I-Application
structure	I-Application
(	O
i.e.	O
,	O
TFSn	O
)	O
.	O
</s>
<s>
The	O
enhancement	O
observed	O
in	O
neural	O
envelope	B-Algorithm
responses	O
is	O
consistent	O
with	O
enhanced	O
auditory	O
perception	O
of	O
modulations	O
following	O
cochlear	O
damage	O
,	O
which	O
is	O
commonly	O
believed	O
to	O
result	O
from	O
loss	O
of	O
cochlear	O
compression	O
that	O
occurs	O
with	O
outer-hair-cell	O
dysfunction	O
due	O
to	O
age	O
or	O
noise	O
overexposure	O
.	O
</s>
<s>
Thus	O
,	O
not	O
surprisingly	O
the	O
relative	O
effects	O
of	O
outer-hair-cell	O
and	O
inner-hair-cell	O
dysfunction	O
have	O
been	O
predicted	O
with	O
modeling	O
to	O
create	O
individual	O
differences	O
in	O
speech	O
intelligibility	O
based	O
on	O
the	O
strength	O
of	O
envelope	B-Algorithm
coding	O
of	O
speech	O
relative	O
to	O
noise	O
.	O
</s>
<s>
For	O
sinusoidal	O
carriers	O
,	O
which	O
have	O
no	O
intrinsic	O
envelope	B-Algorithm
(	O
ENVp	O
)	O
fluctuations	O
,	O
the	O
TMTF	O
is	O
roughly	O
flat	O
for	O
AM	O
rates	O
from	O
10	O
to	O
120Hz	O
,	O
but	O
increases	O
(	O
i.e.	O
</s>
<s>
Provided	O
that	O
the	O
carrier	O
is	O
fully	O
audible	O
,	O
the	O
ability	O
to	O
detect	O
AM	O
is	O
usually	O
not	O
adversely	O
affected	O
by	O
cochlear	O
hearing	O
loss	O
and	O
may	O
sometimes	O
be	O
better	O
than	O
normal	O
,	O
for	O
both	O
noise	O
carriers	O
and	O
sinusoidal	O
carriers	O
,	O
perhaps	O
because	O
loudness	B-Algorithm
recruitment	O
(	O
an	O
abnormally	O
rapid	O
growth	O
of	O
loudness	B-Algorithm
with	O
increasing	O
sound	O
level	O
)	O
“	O
magnifies	O
”	O
the	O
perceived	O
amount	O
of	O
AM	O
(	O
i.e.	O
,	O
ENVn	O
cues	O
)	O
.	O
</s>
<s>
Speech	O
that	O
is	O
processed	O
with	O
noise	O
vocoder	B-Application
such	O
that	O
mainly	O
envelope	B-Algorithm
information	O
is	O
delivered	O
in	O
multiple	O
spectral	O
channels	O
was	O
also	O
used	O
in	O
investigating	O
envelope	B-Algorithm
processing	O
in	O
hearing	O
impairment	O
.	O
</s>
<s>
Here	O
,	O
hearing-impaired	O
individuals	O
could	O
not	O
make	O
use	O
of	O
such	O
envelope	B-Algorithm
information	O
as	O
well	O
as	O
normal-hearing	O
individuals	O
,	O
even	O
after	O
audibility	O
factors	O
were	O
taken	O
into	O
account	O
.	O
</s>
<s>
The	O
perception	O
model	O
of	O
ENV	O
processing	O
that	O
incorporates	O
selective	O
(	O
bandpass	O
)	O
AM	O
filters	O
accounts	O
for	O
many	O
perceptual	O
consequences	O
of	O
cochlear	O
dysfunction	O
including	O
enhanced	O
sensitivity	O
to	O
AM	O
for	O
sinusoidal	O
and	O
noise	O
carriers	O
,	O
abnormal	O
forward	O
masking	O
(	O
the	O
rate	O
of	O
recovery	O
from	O
forward	O
masking	O
being	O
generally	O
slower	O
than	O
normal	O
for	O
impaired	O
listeners	O
)	O
,	O
stronger	O
interference	O
effects	O
between	O
AM	O
and	O
FM	B-Application
and	O
enhanced	O
temporal	O
integration	O
of	O
AM	O
.	O
</s>
<s>
The	O
performance	O
of	O
the	O
hearing-impaired	O
individuals	O
was	O
best	O
captured	O
when	O
the	O
model	O
combined	O
the	O
loss	O
of	O
peripheral	O
amplitude	B-Application
compression	O
resulting	O
from	O
the	O
loss	O
of	O
the	O
active	O
mechanism	O
in	O
the	O
cochlea	O
with	O
an	O
increase	O
in	O
internal	O
noise	O
in	O
the	O
ENVn	O
domain	O
.	O
</s>
<s>
As	O
for	O
behavioral	O
data	O
,	O
six-month-old	O
infants	O
require	O
larger	O
frequency	O
transitions	O
to	O
detect	O
an	O
FM	B-Application
change	O
in	O
a	O
1-kHz	O
tone	O
compared	O
to	O
adults	O
.	O
</s>
<s>
However	O
,	O
4-month-old	O
infants	O
are	O
able	O
to	O
discriminate	O
two	O
different	O
FM	B-Application
sweeps	O
,	O
and	O
they	O
are	O
more	O
sensitive	O
to	O
FM	B-Application
cues	O
swept	O
from	O
150Hz	O
to	O
550Hz	O
than	O
at	O
lower	O
frequencies	O
.	O
</s>
<s>
In	O
school-age	O
children	O
,	O
performance	O
in	O
detecting	O
FM	B-Application
change	O
improves	O
between	O
6	O
and	O
10	O
years	O
and	O
sensitivity	O
to	O
low	O
modulation	O
rate	O
(	O
2Hz	O
)	O
is	O
poor	O
until	O
9	O
years	O
.	O
</s>
<s>
For	O
speech	O
sounds	O
,	O
only	O
one	O
vocoder	B-Application
study	O
has	O
explored	O
the	O
ability	O
of	O
school	O
age	O
children	O
to	O
rely	O
on	O
TFSp	O
cues	O
to	O
detect	O
consonant	O
changes	O
,	O
showing	O
the	O
same	O
abilities	O
for	O
5-years-olds	O
than	O
adults	O
.	O
</s>
<s>
The	O
strength	O
of	O
phase	O
locking	O
to	O
the	O
temporal	B-Application
fine	I-Application
structure	I-Application
of	O
signals	O
(	O
TFSn	O
)	O
in	O
quiet	O
listening	O
conditions	O
remains	O
normal	O
in	O
peripheral	O
single-neuron	O
responses	O
following	O
cochlear	O
hearing	O
loss	O
.	O
</s>
<s>
Several	O
psychophysical	O
studies	O
have	O
shown	O
that	O
older	O
people	O
with	O
normal	O
hearing	O
and	O
people	O
with	O
sensorineural	O
hearing	O
loss	O
often	O
show	O
impaired	O
performance	O
for	O
auditory	O
tasks	O
that	O
are	O
assumed	O
to	O
rely	O
on	O
the	O
ability	O
of	O
the	O
monaural	O
and	O
binaural	O
auditory	O
system	O
to	O
encode	O
and	O
use	O
TFSn	O
cues	O
,	O
such	O
as	O
:	O
discrimination	O
of	O
sound	O
frequency	O
,	O
discrimination	O
of	O
the	O
fundamental	O
frequency	O
of	O
harmonic	O
sounds	O
,	O
detection	O
of	O
FM	B-Application
at	O
rates	O
below	O
5Hz	O
,	O
melody	O
recognition	O
for	O
sequences	O
of	O
pure	O
tones	O
and	O
complex	O
sounds	O
,	O
lateralization	O
and	O
localization	O
of	O
pure	O
tones	O
and	O
complex	O
tones	O
,	O
and	O
segregation	O
of	O
concurrent	O
harmonic	O
sounds	O
(	O
such	O
as	O
speech	O
sounds	O
)	O
.	O
</s>
<s>
To	O
compress	O
the	O
large	O
range	O
of	O
sound	O
levels	O
encountered	O
in	O
everyday	O
life	O
into	O
the	O
small	O
dynamic	O
range	O
of	O
the	O
hearing-impaired	O
person	O
,	O
hearing	O
aids	O
apply	O
amplitude	B-Application
compression	O
,	O
which	O
is	O
also	O
called	O
automatic	O
gain	O
control	O
(	O
AGC	O
)	O
.	O
</s>
<s>
As	O
a	O
result	O
of	O
compressing	O
the	O
level	O
,	O
AGC	O
reduces	O
the	O
amount	O
of	O
envelope	B-Algorithm
fluctuation	O
in	O
the	O
input	O
signal	O
(	O
ENVp	O
)	O
by	O
an	O
amount	O
that	O
depends	O
on	O
the	O
rate	O
of	O
fluctuation	O
and	O
the	O
speed	O
with	O
which	O
the	O
amplification	O
changes	O
in	O
response	O
to	O
changes	O
in	O
input	O
sound	O
level	O
.	O
</s>
<s>
AGC	O
can	O
also	O
change	O
the	O
shape	O
of	O
the	O
envelope	B-Algorithm
of	O
the	O
signal	O
.	O
</s>
<s>
Almost	O
all	O
information	O
in	O
cochlear	O
implants	O
is	O
conveyed	O
by	O
the	O
envelope	B-Algorithm
fluctuations	O
in	O
the	O
different	O
channels	O
.	O
</s>
<s>
The	O
processing	O
in	O
cochlear	O
implants	O
is	O
such	O
that	O
the	O
TFSp	O
is	O
discarded	O
in	O
favor	O
of	O
fixed-rate	O
pulse	O
trains	O
amplitude-modulated	B-Device
by	O
the	O
ENVp	O
within	O
each	O
frequency	O
band	O
.	O
</s>
<s>
Usually	O
,	O
the	O
ENVpof	O
the	O
signal	O
in	O
each	O
channel	O
is	O
transmitted	O
to	O
the	O
implanted	O
electrodes	O
in	O
the	O
form	O
an	O
electrical	O
pulses	O
of	O
fixed	O
rate	O
that	O
are	O
modulated	O
in	O
amplitude	B-Application
or	O
duration	O
.	O
</s>
<s>
In	O
humans	O
,	O
hearing	O
loss	O
may	O
result	O
in	O
an	O
unbalanced	O
representation	O
of	O
speech	O
cues	O
:	O
ENVn	O
cues	O
are	O
enhanced	O
at	O
the	O
cost	O
of	O
TFSn	O
cues	O
(	O
see	O
:	O
Effects	O
of	O
age	O
and	O
hearing	O
loss	O
on	O
temporal	O
envelope	B-Algorithm
processing	O
)	O
.	O
</s>
<s>
The	O
need	O
for	O
such	O
tests	O
is	O
corroborated	O
by	O
strong	O
correlations	O
between	O
slow-FM	O
or	O
spectro-temporal	O
modulation	O
detection	O
thresholds	O
and	O
aided	O
speech	O
intelligibility	O
in	O
competing	O
backgrounds	O
for	O
hearing-impaired	O
persons	O
.	O
</s>
<s>
These	O
changes	O
in	O
the	O
perception	O
of	O
the	O
signal	O
can	O
often	O
be	O
predicted	O
by	O
measuring	O
the	O
associated	O
changes	O
in	O
the	O
signal	O
envelope	B-Algorithm
and/or	O
temporal	B-Application
fine	I-Application
structure	I-Application
(	O
TFS	O
)	O
.	O
</s>
<s>
Changes	O
in	O
the	O
signal	O
envelope	B-Algorithm
can	O
be	O
measured	O
using	O
several	O
different	O
procedures	O
.	O
</s>
<s>
The	O
presence	O
of	O
noise	O
or	O
reverberation	O
will	O
reduce	O
the	O
modulation	O
depth	O
of	O
a	O
signal	O
,	O
and	O
multiband	O
measurement	O
of	O
the	O
envelope	B-Algorithm
modulation	O
depth	O
of	O
the	O
system	O
output	O
is	O
used	O
in	O
the	O
speech	O
transmission	O
index	O
(	O
STI	O
)	O
to	O
estimate	O
intelligibility	O
.	O
</s>
<s>
An	O
extension	O
to	O
the	O
STI	O
estimates	O
the	O
change	O
in	O
modulation	O
by	O
cross-correlating	O
the	O
envelopes	B-Algorithm
of	O
the	O
speech	O
input	O
and	O
output	O
signals	O
.	O
</s>
<s>
A	O
related	O
procedure	O
,	O
also	O
using	O
envelope	B-Algorithm
cross-correlations	O
,	O
is	O
the	O
short-time	O
objective	O
intelligibility	O
(	O
STOI	O
)	O
measure	O
,	O
which	O
works	O
well	O
for	O
its	O
intended	O
application	O
in	O
evaluating	O
noise	O
suppression	O
,	O
but	O
which	O
is	O
less	O
accurate	O
for	O
nonlinear	O
distortion	O
.	O
</s>
<s>
Envelope-based	O
intelligibility	O
metrics	O
have	O
also	O
been	O
derived	O
using	O
modulation	O
filterbanks	O
and	O
using	O
envelope	B-Algorithm
time-frequency	O
modulation	O
patterns	O
.	O
</s>
<s>
Envelope	B-Algorithm
cross-correlation	O
is	O
also	O
used	O
for	O
estimating	O
speech	O
and	O
music	O
quality	O
.	O
</s>
<s>
Envelope	B-Algorithm
and	O
TFS	O
measurements	O
can	O
also	O
be	O
combined	O
to	O
form	O
intelligibility	O
and	O
quality	O
metrics	O
.	O
</s>
<s>
Using	O
a	O
model	O
of	O
the	O
impaired	O
periphery	O
leads	O
to	O
more	O
accurate	O
predictions	O
for	O
hearing-impaired	O
listeners	O
than	O
using	O
a	O
normal-hearing	O
model	O
,	O
and	O
the	O
combined	O
envelope/TFS	O
metric	O
is	O
generally	O
more	O
accurate	O
than	O
a	O
metric	O
that	O
uses	O
envelope	B-Algorithm
modulation	O
alone	O
.	O
</s>
