►
From YouTube: Kickoff (2014 Fall NuPIC Hackathon)
Description
NuPIC Community Flag-Bearer Matt Taylor kicks off the hackathon. Jeff Hawkins shares a few thoughts. Matt also makes some community announcements about recent projects.
A
A
B
Welcome
thanks
everybody
for
coming
to
the
fall,
2014,
newport,
hackathon,
hope
you,
like
our
canadian
theme,
it's
a
fall
theme.
My
name
is
matt
taylor.
If
you
don't
know
me,
I'm
the
flag
bearer
for
the
new
community
so
glad
to
see
new
and
old
faces
in
the
crowd.
B
First
of
all,
I
am
recording
things
so
just
to
get
that
out
of
the
way
through
the
camera
back
there
I'll
probably
be
walking
around
with
it.
At
some
point,
I'm
not
live
streaming,
though.
So
I'll
be
editing
all
these
videos,
and
so
what?
If
something
crazy,
embarrassing
happens?
Don't
worry.
B
I
won't
put
it
in
all
of
our
sessions
and
the
demos
tomorrow
will
be
recorded,
but
you
might
be
on
the
tape
if
you
have
a
problem
with
that,
come
and
talk
to
me
and
I'll,
make
sure
that
I
record
you
all
the
time
please
share
on
social
media.
If
you
take
any
photos
or
videos,
that's
perfectly
fine
use
the
new
hack
and
I'm
kicking
around
the
idea
of
maybe
tomorrow
for
those
of
you
who
are
still
around
that
have
worked
through
the
hackathon.
B
Maybe
doing
a
few
video
testimonials
if
you're
interested
in
doing
that
come
and
talk
to
me.
So
thank
you
to
the
staffers
who
have
helped
me
out
with
this
event.
Obviously
we
have
mr
jeff
hawkins
super
taiya,
then
some
other
new
pick
engineers,
scott
purdy,
austin
marshall,
myself,
jake
server.
I
miss
anyone,
I'm
not
sure,
and
then
we
have
some
support
staff.
Terry
fry
is
our
office
manager
she's
extremely
helpful.
In
these
events,
taylor
works
is
also
helping
out
she's
honored
to
meet
the
staff.
B
My
wife
trinity,
taylor
is
also
helping
out,
so
thank
you.
Honey,
kyle
quackenbush
is
one
of
taylor's
friends
who's,
also
helping
and
a
big
thank
you
to
pinger.
Once
again,
this
is
their
space.
They
gave
it
to
us
to
use
for
the
weekend.
That's
very
nice
of
them.
They
pretty
much
gave
us
free
reign
to
the
whole
space
and
just
said
bye.
I
love
this
long,
so
be
respectful.
B
I
would
like
to
bring
jeff
up
just
for
a
moment
to
say
a
couple
of
words
to
start
off
the
hackathon.
Just
because
he's
such
an
inspirational.
D
So
a
lot
of
us
just
came
off
this
workshop.
This
day
was
a
lot
of
work
putting
together
and
I
think
it
went
out
pretty
well.
We
had
some
of
100
people
in
the
school
of
cool
courtroom
in
the
city
and
we
had
to
put
these
presentations
together.
So
some.
E
D
Us
kind
of
like
wasted
from
that
those
are
emotionally
draining
mentally
training
events,
so
maybe
go
to
a
slower
start
this
morning
than
normal,
but
it
was
really
kind
of
fun.
It
was
cool.
We
had
people
from
actually
around
the
world
showed
up
that
I
couldn't
believe
and-
and
it
was
kind
of
fun
to
sort
of
present
the
state
of
the
art
of
our
research
and
but
I
think
you
know
I
just
want
to
thank
thank
you
for
being
here.
D
I
think
the
hackathons
are
a
really
nice
part
of
what
we
do.
The
whole
loop
of
community
actually
is
really
nice
part
of
what
we
do
as
you've
heard
me
say
many
times,
probably
not
sitting
many
times.
Everybody
says
many
times
and
no
mention.
We
don't
view
this
as
sort
of
our
thing.
This
is
like
hey
we're
trying
to
build
the
movements
and
finding
people
to
do
this
stuff.
Trying
to
get
people
interested
in
cortical
models
show
businesses.
D
D
So
I
met
yesterday
because
you
had
this
picture
of
the
homebrew
computer
club
in
this
presentation.
You're
saying
I
feel
a
little
bit
like
that.
Every
time
I
come
here,
I
have
no
idea
what
hungry
is
really
like.
D
Maybe
the
hackathon
there's
no
big
factory
talks
these
days,
so
I
really
have
nothing
else
further
to
say
we're
gonna
be
around
here.
Some
of
us
are
doing
hacks.
I
don't
know.
I
don't
want
to
talk
anymore,
but
we're
gonna
have
some
discussions
I'll
be
doing
tomorrow's
email
emails
while
doing
some
sort
of
a
q
a
session
with
a
presentation,
and
that's
it.
Thank
you
being
here.
I
have
no
great
reason
to
depart
this
morning,
boy
ever
actually,
but
I'm
happy
to
I'm
happy
you're
all
here.
Thanks
guys
thanks
so
much.
B
Okay,
a
bit
of
hackathon
protocol.
First
of
all,
the
place
to
go
for
details
is
convention.org
hack,
I'll.
I
have
a
I'll
have
a
tv
in
the
back,
that's
kind
of
displaying
any
announcements
or
either
that
or
the
registered
hacks
which
I'll
talk
about
in
a
minute
the
code
of
conduct.
Actually,
let
me
just
go
through
here,
so
you
can
see
it.
No
okay,
nevermind.
We
also
have
a
code
of
conduct.
B
B
We're
not
okay
whatever,
so
our
code
of
contact
basically
says
you're
not
going
to
be
offensive.
You're
you're
not
going
to
do
anything
that
is
aggressive
or
offensive
to
any
group.
So
please
don't
we
definitely
don't
want
that.
There
is
a
place
for
you
to
register
your
act.
That
was
what
I
was
trying
to
get
to.
B
Which
is
not
currently
loading,
but
if
you
go
to
demented.org
hack,
there's
a
link
to
hack
registration
and
it's
just
a
google
form
where
you
can
put
the
name
of
your
hack,
any
members
that
you
have
in
your
group.
I
encourage
you
to
try
and
break
into
groups,
you're
actually
going
to
be
easier
if
you
break
up
the
groups,
and
so
you
need
to
register
your
hack
in
order
for
me
to
get
a
good
idea
of
how
many
demos
we're
going
to
have
tomorrow.
B
So
I'm
going
to
be
reminding
you
of
that
walking
around
if
you're
working
on
something
making
sure
that
you
have
registered
I'll
go
over
the
schedule.
Real
quick,
but
it's
also
posted
on
that
website,
make
a
few
community
announcements
and
I
should
have
removed
that
last
one.
B
So
the
schedule
today
is
as
soon
as
this
is
over.
If
you
are
having
problems
getting
new
pic
running
on
your
computer,
go
into
this
conference
room
right
here,
it's
called
the
fishbowl,
apparently
so
go
into
the
fish
bowl
and
we'll
have
a
couple
of
nimenta
engineers
in
there
to
try
and
help
you
get
nuked
running.
B
Maybe
10.?
Okay,
so
if
have
you
tried
it
and
had
problems?
Okay,
so
maybe
it's
easy.
Maybe
it's
not!
Let's
meet
up
in
the
fishbowl
after
this
meeting.
If
you
want
some
help
getting
it
installed,
so
we'll
try
and
spend
half
an
hour
so
getting
our
engineers
helping
you
at
one
o'clock,
francisco
weber
from
cortical,
I
o
will
give
us
a
presentation
on
the
way
to
language
intelligence,
about
their
developments
with
their
products,
which
is
a
very
exciting
stuff
at
2
30
chain
surfer,
we'll
talk
about
applications
of
htm.
B
If
you're
at
the
workshop,
it's
going
to
be
basically
the
same
presentation,
but
he'll
have
more
time
to
go
into
more
detail.
If
you
have
questions
so,
if
you're
at
the
workshop,
you
may
want
to
skip
that
tomorrow,
jeff
is
going
to
be
back
at
2
p.m.
And
if
you
remember
at
our
last
hackathon,
we
did
kind
of
something
I
called
it.
A
fireside
chat
with
jeff
and
we
just
brought
out
a
whiteboard
and
people
huddled
around
and
jeff
talked
about.
B
The
new
temporal
memory
features
that
we're
trying
to
implement,
so
I
want
to
do
something
similar
to
that.
I
don't
think
jeff
has
anything
groundbreaking
too
new
to
talk
about
we're
we're
trying
to
implement
some
some
of
the
stuff
we
talked
about
last
time,
but
I
liked
the
format
of
getting
jeff
in
front
of
a
whiteboard
with
a
bunch
of
geeks
asking
them
questions,
so
we're
going
to
force
jeff
through
that
again
and
at
four
o'clock
tomorrow
we're
going
to
start
backcom
demos.
B
I
don't
know
how
long
it's
going
to
last
last
time
it
was
probably
three
hours
so,
depending
on
how
many
demos
we
have.
I
may
time
limit
you
and
I'll
get
I'll.
B
B
This
map
is
a
somewhat
up
to
date,
but
this
is
the
upstairs
of
the
building
that
we're
in.
I
have
some
work
stations
set
up
upstairs
if
you
see
any
long
tables
with
tablecloths
and
chairs
set
up
to
them
and
there's
power
strips
underneath,
that's
a
great
place
to
work.
So
there's
some
along
the
back
wall
here
here,
there's
one
overview
right
behind
the
presentation
area.
There
are
also,
if
you
go
downstairs.
Oh
these
are
the
emergency
exits,
but
I
have
no
confidence
that
the
doors
are
unlocked
so
you're
on
here.
B
Downstairs
we've
got
the
main
kitchen,
which
is
hard
to
miss
and
for
the
most
part,
the
kathy
at
pinger
did
a
great
job
walking
around
and
putting
tags
on.
Things
like
this
room
is
okay
to
use.
You
know,
because
the
sign
says,
okay
to
you,.
B
B
Downstairs
there
are
some
workstations
if
you
are
a
lone
wolf
type
and
you
don't
want
to
be
elbow
to
elbow
with
anyone.
There
are.
B
That
you,
you
might
find
that
are
just
kind
of
one-off
work
stations,
there's
not
a
whole
lot,
there's,
maybe
four
or
five.
So
if
you're,
that
type
of
person,
you
should
go
claim
it
otherwise
you'll
you
can
work
in
any
one
of
these
rooms
to
say,
okay
to
use,
you
can
just
park
and
work
wherever
you
like.
B
Guess
we're
not
going
to
get
the
projector
back
on
okay,
so
I'll
just
go
through
the
rest
of
the
pro
call.
Without
slicing
this
facility,
you
can
come
and
go
as
you
please,
but
somebody
has
to
be
downstairs
to
let
you
back
in.
So
there
is
a.
E
B
Here,
if
you
like,
you
don't
have
to
you're
under
no
obligation.
I
don't
pressure
anyone
to
work
too
hard,
but
I'll
probably
be
here
most
of
the
night
and
a
few
other
people
I
know,
tend
to
stay.
B
Some
something
weird
going
on
with
everything
right
now.
I
do
have
some
sleeping
mats
that
we
rented
from
rei
if
you're
a
camper
type
so
and
there's
a
bunch
of
pillows
through
here,
so
you
can
just
do
what
you
can
to
make
yourself
comfortable.
If
you
find
yourself
very
tired
and
you
need
sleep,
otherwise
feel
free
to
go
home
or
go
to
hotel
or
whatever
and
come
back,
but
remember.
I
think
there
is
a
10
o'clock
presentation
tomorrow
morning.
If
you
want
to
no
no
2
o'clock,
that's
jeff's
whiteboard
session.
G
Also
guys,
if
your
hack
is
using
real-time
connections,
you
might
want
to
think
about
using
your
phone
for
the
presentation,
because
the
local
wi-fi
sometimes.
A
B
Shouldn't
be
a
problem
this
time
so
we'll
find
out.
Okay
remaining
protocol
wear
your
badge
around
please.
I
know
it
might
get
uncomfortable,
but
it
helps
me
know
what
your
name
is.
I
want
to
try
and
find
people's
names.
We
will
sort
of
walk
around
and
announce
when
food
arrives.
I
think
there's
breakfast
downstairs
right
now,
which
is
like
fruit
and
pastries
and
stuff,
we'll.
B
So
just
wander
down,
it
seems
like
last
year
last
time,
there's
always
food
laying
around
in
the
kitchen.
If
you're,
hungry,
try
and
pick
up
after
yourselves,
please
it
makes
my
job
a
lot
easier
at
the
end
of
this
hackathon.
So
if
you
see
trash
laying
around
and
you
see
a
trash
basket
like
right,
there,
just
put
it
in
yeah-
please
be
respectful
of
each
other.
We
haven't.
H
H
And
please
read.
B
B
Of
context,
okay,
so
for
hackathon
communication
we
not
only
have
an
irc
channel.
At
this
point,
we
also
are
using
this
tool
called
getter,
which
is
like
github
chat,
which
I
found
really
nice,
so
I'm
usually
on
it,
I'm
also
on
irc,
so
you
so.
I
try
not
to
miss
out
on
anything
but
there's
a
public
room
on
getter
that
you
can
get
to.
You
do
have
to
authenticate
github.
So
if
you
have
a
github
id,
you
can
get
on
twitter
pretty
easily
and.
B
And
please
feel
free
to
use
the
mailing
list
or
the
mailing
list
is
our
resource.
We
can
use
it.
However,
we
want
so
email
to
new
quick
discuss
if
you've
got
technical
questions
and.
B
The
list
as
well
as
I
can,
and
maybe
if
you
can't
find
an
answer
in
the
room,
we
can
find
an
answer
on
the
mailing
list.
B
A
lot
of
people
interested
in
what's
going
on
with
the
hackathon,
while
they're
at
home,
we
do
have
a
place
where
you
can
report
bugs
on
demento.org
there's
a
reporter
bug
link
on
the
menu.
So
if
you
do
find
a
bug
or
you
find
a
or
you
have
problems
doing
something,
you
think
should
be
simple-
just
report
it
as
a
bug
so
that
we
can
try
and
address
it.
B
Our
hashtag
is
new
pickpack.
Like
I
mentioned,
let's,
let's
try
and
use
social
media
if
you
use
social
media,
please
please
spread
the
word
about
the
icon.
B
Okay,
I
want
to
make
a
few
quick
announcements
and
I'm
getting
out
of
your
hair,
so
you
can
start
get
going.
One
of
the
things
we've
done
recently
is
we
created
a
new
github
organization
called
new
big
community.
This
is
not
a
new
menta
initiative.
This
is
a
community
initiative.
This
is
something
the
community
wanted.
They
wanted
a
place
where
we
could.
We
could
collect
new
pick
projects
and
that's
what
this
is.
B
So
we
currently
contains
virgo
burns
project
vortex
and
david
rigozzi's
project
new
fix
studio
and
jim
bridgewater,
pro
jim
bridge,
roger
jim
bridgewater's
project,
nuked
vision,
which
was
susan,
deepik
project.
I've
thrown
a
couple
things
up
there
that
that
I've
been
working
on
for
better
community
initiatives
if
you
create
a
hack-
and
you
want
to
put
your
code
up
here-
let
me
know
I'd
be
happy
to
collect
it
in
this
space
and
there's
there's
other
cla
projects
on
our
wiki
and
some
of
those
might
end
up
there
as
well.
B
B
A
visualization
tool,
but
it's
kind
of
a
way
to
construct
hierarchies
and
share
it
among
members
of
the
community
caveats
I
I
haven't
had
any
time
to
really
look
into
this
project.
It
looks
cool
but
david's,
really
dedicated
to
it
and
there's
several
community
members
that
are
that
are
actively
involved
in
it
now
testing
out
and
stuff.
So
you
can
do
a
hack
based
on
this.
B
Another
community
project
is
david,
ray
has
done
a
complete
job
report
of
spatial
pooling
and
the
new
temporal
memory
algorithms.
So
I
think
this
is
cool
and
he's
he
came
to
us
and
said
I
I
just
want
to
give
this
to
you
guys
so
he's.
B
A
B
Been
he's
been
running
on
this
and
working
on
it
for
a
few
months
now
and
he's
doing
a
really
great
job
he's
got
a
one-to-one
correlation
between
his
methods
and
functions
and
the
python
version.
So
he
tried
to
stick
to
the
python
api
and
he
estimates
that
98
of
the
python
tests
in
each
module
have
the
exact
same
output
produced
within
the
java
unit
test.
So
that's
pretty
impressive.
He's
got
the
algorithms
down
to
a
t.
B
B
Get
get
some
answers
from
you
guys
who
has
ever
been
to
a
hackathon,
okay,
who's
ever
built
new
pick.
I
know
we
already
did
some
of
this
stuff
hands.
B
Come
on
built
new
pick,
okay,
there's
a
few
down
so
you're
the
guys
that
need
probably
installation,
help
ones
that
don't
have
your
hands
up.
Have
you
ever
run
any
of
our
example
code?
E
B
If
you
have
any
ideas
for
improving
this,
let
me
know-
and
I
will,
but
this
will
kind
of
walk
you
through
the
steps
of
of
how
to
get
productive
with
newpick
and
now
is
the
time
to
talk
about
ideas,
so
hack
ideas.
B
So
there's
one
thing:
I
know
people
are
interested
in
doing
eeg
stuff.
I've
done
some
eeg.
I've
tried
to
do
some
eeg
stuff.
We
had
a
season
of
new
pick
project
that
several
of
us
got
together
and
tried
to
do
eeg
stuff
and
we
we
got
to
a
certain
point
and
we
got
blocked
and
we
couldn't
move
any
further.
B
Who
is
interested
in
doing
eeg,
parsing
or
eeg
analysis
with
new
pick?
Hence
high.
There's
two
three
four
guys
look
around
and
and
spot
these
keep
your
hands
up,
keep
your
hands
up
and
look
around.
I
want
you
guys
to
connect
if
anybody
is
interested
in
this.
This
might
be
a
good
team
to
put
together
share
some
ideas,
because
it's
hard,
it's
not
it's
not
as
easy
as
I
thought
it
was
going
to
be
the
eeg
data
that
we
got
for
the
kaggle.
B
So
we
have
some
knowledge
here
that
we
can
share
and
let's
try
and
do
that,
there's
also
an
idea.
Let's
see
if
this
breaks
my
machine
again,
there's.
B
On
this
wiki
page,
one
of
them
was
creating
some
type
of
tool
to.
I
think
it
was
for
model,
parameter
selection
tool
and
that's
not
gonna
work.
So
if
you.
B
I
would
like
to
do
a
minecraft
hack.
I
don't
know
if
anybody's
interested
in
that
has
anybody
created
any
minecraft
mods
or
so
I
have
an
idea.
I
want
to
take
the
geospatial
encoder
and
I
want
to
feed
it.
Xyz
coordinates
from
minecraft
for
all
of
the
moving
entities
in
the
game
and
get
anomaly
indications
and
display
them
in
the
game
so
that,
as
I'm
walking
around,
I
can
see
where
newpick
thinks.
I'm
going.
That's
my
idea.
If
anybody's
interested
in
that
come.
B
B
J
Framework-
and
in
fact
you
know
the
framework
I
have
interrupted
somehow
but
I'd
like
to
automate
or
to
semi-automate
at
least
the
use
of
some
of
the
new
big
functions.
J
B
J
I'd
like
to
select
a
few
of
the
functions,
I'd
like
to
start
with
the
one
of
the
samples
you
use
to
demonstrate
this
and
I'd
like
to
take
some
of
the
functions
of
newbig
from
there.
I'd
like,
on
top
of
them
to
put
a
sort
of
graph
queries
and
those
graph
queries.
I
want
to
match
them
with
the
problem
context
you
use
in
that
sample,
so
basically
it
will
automatically
apply.
It
will
rebuild
the
code
of
the
sample,
but
basically
automatically
similar
to
it.
B
E
Hey
everyone,
I'm
rhian,
and
so
there
are
two
ideas
that
I'm
really
interested
in
exploring.
The
first
is
to
be
able
to
use
hdm
to
either
predict
or
detect
anomalies
in
network
structures,
so
by
network
structures,
I'm
essentially
meaning
I'm
referring
to
data
structures
in
which
nodes
represent
distinct
elements
and
links
or
edges,
represent
connections
between
networks.
E
So
just
typical
examples
of
networks,
social
networks,
biological
networks,
protein
networks,
gene
regulatory
networks,
citation
networks,
yeah
I
mean
there
are
a
lot
of
potential
cool
applications
and
the
other
idea
I'm
actually
really
interested
in
functional
programming
and
lisp
stuff.
So
if
anyone
is
doing
any
sort
of
list
or
functional
programming,
I
know
we
have
cortex,
so
I'd
definitely
be
interested
in
talking
to
you
at
some
point:
that's
it
cool!
Thank
you.
Thanks.
G
Occasionally,
you
analyze
data
that
comes
in
repeated
sequences,
say
you
analyze
soccer
matches,
which
is
a
set
of
sequences
in
time,
but
they're
all
kind
of
similar,
and
if
you're
analyzing
a
lot
of
them,
it
would
make
sense
for
you
to
use
a
lot
of
different
machines
at
once.
So
I'm
thinking
about
writing
a
way
to
use
new
pick
in
a
distributed
environment,
to
kind
of
converge,
the
learning
into
the
same
model,
but
be
able
to
run
multiple
instances
of
music
at
once,
more
working
on
the
same
problem,
cool.
I
Hey
guys,
I
run
a
group
of
books
on
the
apple
watch,
and
so
I'm
interested
in
thinking
about
what
kinds
of
applications
would
be
good
for
a
wrist
wearable,
and
so
as
an
example,
you
know
you
can
get
a
user's
heartbeat
or
you
know.
Perhaps
you
can
tie
into
some
by
api,
so
we're
giving
them
some
kind
of
predictive
data
through
that.
I
B
B
Actually
this
is
not
doing
audio
analysis,
experiments
so
there's.
B
H
What
I'm
interested
in
is
not
so
much
about
anomaly,
but
to
look
at
you
know,
predictive
patterns
and
and
to
see
what
is
the
pattern
within
the
pattern
right,
so
the
ten
scientists
are
changing
it,
which
are
the
10
factors,
are
actually
resulting
in
a
predictive
pattern
so
to
identify
pythons
within
data,
not
so
much
interested
in
anonymities,
but
just
to
understand.
Why
is
a
pattern
operating.
A
B
C
C
My
colleague
mark
we
are
interested
in
a
natural
language
processing
problem,
we're
going
to
try
to
train
a
network
with
fingerprints
that
we're
deriving
from
the
word
net
semantic
dictionary
and
then
use
that
to
predict
sequences
of
meaning
and
feed
that
into
a
sentence
generator
to
generate
grammatical
but
meaningful
sentences
after
having
trained
it
off
of
some
text
corpuses.
C
We've
made
a
little
bit
of
a
start
on
it,
but
hopefully,
by
the
end
of
the
weekend,
we'll
have
sentences
coming
out
that
are
based
on
meanings
that
the
system
has
done.
Are
you
familiar
with
what
portugal
I
o?
Yes,
okay,
we
had
a
long
talk
last
night
about
and
there's
some
differences
in
the
approaches
we're
taking
to
generating
this
fingerprint
so
it'll
be
interesting
to
see
okay.
Well,
what
comes
out
of
that
nice.
K
Group
up
in
some
teams,
hi
everyone,
so
at
numenta
we've
been
doing
research
on
sensory
motor
inference
and
terminal
cooling,
it's
still
very
early,
but
you
and
I
will
be
working
on
a
application
of
that
on
robotics.
So
we
have
a
robot
that
moves
on
a
it's.
An
ir
sensor
that
rotates
on
a
platform
and
the
rotation
can
be
programmatically
controlled.
K
So
we'll
set
up
like
a
world
around
it
made
of
cardboard
sheets
at
different
distances
and
have
it
move
around
randomly,
learn
the
sensory
motor
sequence
and
try
to
and
have
it
put
in
different
worlds
and
how
to
see
if
it
can
build
a
stable
representation
of
a
particular
world
and
recognize
which
world
it's
in
so
we'll
be
working
on
this.
B
Registered
my
hack
last
night
before
it
was
even
available
to
you
guys,
but
here's
the
form
so
please
fill
this
out
when
you
are
ready
to
get
an
idea.
Sooner
is
better.
B
Who
did
the
job
report
a
parameter
optimizer?
So
if
you
give
it
an
input,
vector
size
and
column
dimensions,
it
would
give
you
some
same
variable
settings
for
the
attributes
of
spatial
cooler,
temporal
memory
and
stuff
like
that,
so
that.
A
B
F
You
could
try
to
find
surprise
in
music
in
a
level
of
like,
if
you
take
a
sequence,
you
can
download
a
midi
sequence
algorithm
for
music
and
then
possibly
go
to
find
patterns
during
the
music
and
then
go
oh
okay.
Well,
here
is
a
here's,
a
pattern
at
different
time
scales.
So
you
go.
Oh,
this
is
kind
of
like
a
repeat
of
a
phrase.
An
earlier
phrase:
that'd
be
kind
of
an
interesting
application
to
analyze
the
style
of
music.