Saturday, July 5, 2014

  |  No comments  |  

Google Announces Automatic Captions on YouTube

alright good
morning everyone my name is so Jonas
Clank
and I'm the product manager I'm for the
excess built a team here at Google
and I'm where very very excited to have
you guys here today
I want to extend ever really really warm
welcome
and thank you all for for coming here
I'm I know a lot of your local
bite still are early in the morning and
I'm
where thrilled to have you here I think
we're I'm
gonna have a a great event today and I'm
I look forward to showing you the demo
that we have going today
am me we have an agenda here
behind me after I do their brief
introduction
that I'm and blocking I'm
I'm gonna handed over 22 Venser where
then gonna have
I'm can Naomi do some demos
and been to come back to provided rapper
and then
will do some Q&A towards the end I'm the
book covers the majority will be
available afterwards to give some
additional demos
answer your questions anything you
haven't gone into him so
where really thrilled to have been
served with us his title here
Google as vice president and chief
internet evangelist
I'm and but as many as you know he's is
a lot more than that he is a
widely known as the father of the
internet I'm has one of the car
designers
if the TCP IP protocols the basic
architecture
of the Internet he to us
I'm which might be widely known he's I'm
really really critical
as an and favoring supporter for am and
accessible Internet
and especially for those who have ass
special access needs
so what up for the deal we're really
thrilled to have him here
and I give you the answer thank you all
very very much for coming
and welcome to Google's DC office
it's these kinds of events that have
made us so happy to have this facility
here
and I hope during the course of 2010
we'll see you back here
for other kinds of announcements related
to our work
inaccessibility let me start out by
I'm standing in front of these darn
slides how can we do this are not in the
way
as I for offside how's that
okay can I get the first fight please
there we are
everyone should know by this time that
Google's
motto is to organize the world's
information
and make it accessible and useful and
accessible is a really important word in
the context of what we're going to talk
about today
accessibility comes in a wide range
features they are issues for people who
are deaf or hearing-impaired or people
who have
visual I impairments who have low vision
or no vision
people who have motor problems and
things like that Google believes that
the world's information
should be accessible to everyone in our
interest is in finding ways
to achieve that objective we're far from
solving the problem completely
but today's discussion and announcements
will show you what I consider to be an
important step
in that direction I'm gonna cheat look
at my notes on
and I have to use my damn classes to do
it I hate to admit they have to wear
these things
inside every 66-year-old is a
16-year-old wondering what happened
so
I i want to tell you first ball why
accessibility is personally important to
me
and secret who is in the audience over
there wave your hand sacred
and I are both hearing-impaired at
secret was totally deaf for fifty years
she now has to go clear implants and
they work wonderfully well
they work so well we had to buy a bigger
house because she wanted bigger parties
because she can hear
so this is a technology which
spectacular
I've been wearing hearing aids since I
was 13 you can do the math that's fifty
three years
so both of us care a great deal about
how technology can help people with
various impairments get access to
information and be connected
with the rest of the world so quite
apart from my job at Google
I have great personal interest in what
we're talking about today
you know you too is another part
at the Google family and you saw the
signs I hope it said Google and YouTube
are hosting this meeting
you too uncovered an extraordinary
desire
by the world's population to express
itself
using the video medium and one of the
big challenges that the video medium
is whether it can be made accessible to
everyone
it's a little hard to make video
accessible to people who were born to
you need to have
the some kind of descriptive ability
there but for people who were doused
thats and another important area the
concern because they may not be able to
hear people speaking
so videos turned out to be is stunningly
important new medium for personal
expression
I you may not know the statistics but
there are 23
hours video uploaded in to YouTube
every minute 23 hours a video per minute
now I have no idea how many hours a
video or actually being watched but I
think the answer is bigger than that
although a for some videos it may be
like blogs
the average number readers have a blog
is probably 1.1
the person that wrote the blog and his
dog
but these videos are often extremely
popular
and if you haven't noticed they have
become an important medium a political
expression
I you saw this yet very evidently
in the post Iran elections where many
people use
video and you too in particular to tell
the rest of the world what was going on
yeah in that country well we started
our life at Google as a company that was
indexing the Internet we crawl through
all the billions to pages on the net
and we try to index them and then help
people
find the information that they're
interested in we'd like to extend that
ability
to find information to other media
besides texting so part
the what we are about at Google is
finding
other ways indexing content that may not
necessarily be purely textual
in nature we will we move very very
quickly when it comes to new
ideas and we usually get them out there
early so
beta testing is a very common thing for
us we put new products out there
and find out what people do with them an
example this is Google Wave
where we put that out very very early in
its development and we're discovering
very interesting things about how people
use this
combined communication tool which
incorporates blogging
instant messaging email and many other
forms a the interaction
we're very interested in picking up
other kinds have contacts
content than tax so you see this with
Google Earth
you see it with Google Maps I and you
see it with attempts to allow people to
express
information or display information it's
geographically indexed
by using our Google Earth and Google
Maps
facilities we're looking for all kinds
of ways
improving people's ability to find
information so
some love you may be using the Google
Toolbar which
we put their make available so that you
can easily
invoke a Google search we're also very
interested
in internationalization we recently
announced that we're
capable a searching and translating
51 different languages now there's some
linguists in the crowd I'm going to pick
on my good friend Marco prince who has a
wonderful linguist at the National
captioning Institute Mart
I know you heard this a million times
but I have to tell everybody that mark
is the inventor
cling on language a and
yeah he's a he's a linguistic and I'm
sure he can appreciate the translating
51 languages in 251 other languages and
non-trivial exercise
out I now also have to admit that no it
we don't do equally well with all
language pairs
I tried Chinese to Danish while I was in
Copenhagen
earlier this week and since they don't
speak either language I don't know how
well or poorly I did but I noticed there
were some chuckles in the Danish
speaking audience
but the point here is that we recognize
accessibility
as being another way I accessibility by
way of language
as being and other important crane ugh
accessibility
so what we're seeing here are some
statistics basically
relatives cystic save the number of
people who
speak various languages who were on the
Internet
but you'll notice that the blue colored
bars
illustrate people who have accessibility
challenges
so there are more people with problems
have color vision
then there are Japanese or Spanish or
German or French are the other
language speakers or people with poor
vision exceed
oh and represent a a significant part
the internet user population or people
with poor test dexterity or people who
are doubt
or hard of hearing and blind so what
we're saying here with these data
is that there are insignificant number
of people
in the internet population who will
benefit
from improvements in accessibility and
we recognize this on a global scale
we tried to serve users or 1.7 billion
of them who are online
on the Internet today and we would like
to improve
the ability or users who have
disabilities to get access to the
information that we can provide
so the let me I'm the
just give you a couple examples %uh the
things that we deal in Google
applications are Google Labs
we've been focusing on keyboard access
Andy the USAID in for SSD assistive
technologies like screen readers
to make the information available more
accessible to people who
can't see we've released and Android
operating system for mobile Isles in our
intent was to keep that as open as
possible
is an open-source operating system
people can
at functionality to it and one other
things we would like to facilitate
is the addition assistive methods to
make the operating system
more useful in any mobile context
for people who need help we also
released
a browser called chrome and in addition
to its security properties
which we believe are superior it is also
an open platform
one in which existed mechanisms can be
incorporated
and finally YouTube itself I
in 2006 added the ability to put
captions
up on the YouTube videos if the supplier
about video had captioning capable
captioning transcript available and
could
insert them into the video so we're very
interested in helping people
with that kind of guy information to
make it
easily accessible to the rest of the
world so I'm going to
stop here and just remind you again
that Google is fully are prepared
to pursue accessibility features in all
of our products and services
and we're here to tell you about some
specific ones today and to do that
I want to introduce my good friend came
here in steam
can I have known each other for probably
over thirty years
I think we were two or three years old
at the time when we met
can was that SRI International on the
west coast
he and I worked on something called
definite which is an extremely early
attempt
to bring terminal access to the Internet
the two people who were using TTY isn't
things to that kind
I would say we had mixed results with
that experiment
but we learned a lot from how much value
there came
from having access to the internet
through these tools
but in the meantime can has joined
Google
and above all the people
that I know it at Google
can has been the strongest proponent and
one of the strongest
technical contributors to our are s
Accessibility Initiative she's very very
vocal and passionate on the subject
and he's going to show you what he and
his colleagues have been able to do
in this area so can thank you so much
for coming here to the east coast
let me turn the microphone over to you
thank you then hi everyone
my name is Ken here in staying as you
can see
I'm a software engineer and I work for
Google
as always such a great honour to share
the stage
here with that that for me
personally it even a bigger honor to
have all of you here
because I know for all of us we work
many many years
diligently pursuing the same goal
we share the same frustrations and today
I'm really happy to be able to share
something that we hope that we'll reach
this goal
accessibility something report on hard
send some love you probably don't know a
lot about you too
and the captioning services that we can
do
I'd like to start by reviewing quickly
sum up what we already provide
and then we'll talk about the problems
that we're facing and lastly will talk
about power trying to solve those
problems
first all-star little demonstration
this video is one we made when we first
launched
captions on YouTube has really fun
making a
so I may play some
as you can see I'm changing the size
people
wanna see sometimes more video than they
wanna see captions
but if you're in the back you may wanna
seem a little bit bigger
several
all explain the details later how to do
it hun
the also for some people
who wanna see more the video you can
turn off the background
that's an option
most people like the black background as
it makes it easier to read so I'll leave
it on that way
can you read this or I desire in
for everyone he said looking back there
huh okay
let me play a little then
good
lose only sweetie cops on credit you put
it got some common walked in a bonfire
not for long
I V disrepute trip he speaking French
going on the program in america
with a home-ice choice counter to the
shocking videos are you human
and potato gun the polish on them or not
she mention abide by my popcorn
a way of asking other clinical up bro
say about
now Japanese is being spoken
the moment
this person here his name is he wrote
I'll
he's our product manager he's done a
great job
unfortunately he can come here today
that his your own video
so what it is demonstrated is how
captions
have made videos accessible to everyone
doesn't matter if you can hear
or not you know it for people who don't
understand Japanese
the rest and sign languages are made
accessible through captions
now suppose English is not your first
language
this video I'll has several passions
that are already uploaded so I'm able to
take another language
up is this one
has anyone speak both
it a lot from across Canada bag with
bloody kiss in
now suppose
and out Lister
prepared captions you don't see the
language that you want
as bench then mentioned are ready we
support 51 languages
so it's possible took translate to any
one of those other languages
summer better than the others that let
me show you
see this one it says translate captions
so I think that now it's translating
from
English to let's pick one normally I
like to ask people to
help me pick but I know there's a few
people who do speak friend simon did go
ahead and use that
but I have a consul de
you choose all time machine did stone
you up fanciful
nagy mark got sisal in the villages had
a nice
topped a poll on dog day
se que tu no has dormido me miss a deal
to look at it he still speaking Japanese
that we can read it breaks
those things
are wonderful for accessibility but
there's another very cool benefit of
having captions
because they're taxed we can search on
them
there's two things that we can do with
that
and helps you find the video you're
looking for and secondly
it helps you find the exact text you
want within the video
let me show you that it's called a
snippet and you can jump directly to the
snippet
within the video that you're searching
for cement demonstrate that next
here we see
the page from Google is an advanced
video search page
I've already put in the phrase that I
want to search for
ones that's one small step this is one
of my favorites
I've also checked this box
search only closed caption videos so
weak
checked out parts so here we go
and found several
ahead happens to know and i won
now you'll notice it found snippets
and you know that
by offering this option start playing at
search term
so I'll click that and it'll drop me
right there
well actually it jumps to a place just a
little bit before
because it knows that you want a little
bit of context first
still slightly long but I hope it's okay
today
the Google Lunar X PRIZE is challenging
free enterprise to reach much further
to the leave the way to harnessing Beach
Road the
days away its birds corner an
to world system
a crime of crap are for sure
video it may not be that important but
if you have a really long video
this disability is very important
this is one of my favorite teachers
and remember I actually have the
opportunity watch this live
on TV years ago because at that time
there were no captions
as you know but now I get to see it
those things are all there and they have
been there
in YouTube
since we first launched captions
more than a year ago I'm sorry 2008 not
not sex I were very surprised and
pleased at how many
videos have captions and there's a few
hundred thousand videos
that are captioned with a new to you and
we think that's a great number
so what do you think the problem is
let me show you the
can be Shekoni
let me explain your I tried to figure
out a way to
explain without including on a math but
this problem is that we're encountering
so I came up with the visualization you
know the network
is built with a buncha pipes the
Internet is built with a bunch of hype
it's true
and a lot of water runs through these
internet pipes
and a are really video okay I happen to
have some %uh that video
year okay
this is special water it happens to be
captioned
captioned video as you see that water
has the bottle has
words on it caption it's prepared for
you
but
even need you to do me
this is our problem remember what dent
said earlier
every minute we stand here and talk
people are uploading 22-23 hours
a video not minute
hours not 23 videos and South were
talking
hours so Hines and that every minute
every day every month just
it's coming in Co
30 okay
cell do you see any bottles there
can't see any and fortunately
up all that content being uploaded
just small amount is captioned
very small Andy
so the question is
who's gonna bottle out water at Google
we call that a problem scale how to
scale
fortunately at Google and new to
are leaders are very passionate and want
to support
those solutions that addressed scale
maybe you're not familiar with that word
but
our CEO his name is Eric
SMent he loves to talk about this issue
and he's a very clear speakers I would
like to borrow some of his words
tell your then it's time it's time for
us
to take advantage up the amazing
opportunity
that is before us and that's what I want
to talk do you notice anything different
about these captions
mmm I mean this is our new minutes
today and tomorrow you're gonna have
interesting product announcements some
fun stuff and perhaps most important
you're gonna spend time
with the best programmers in the world
who are here in a row
do you notice anything different
there's some mistakes what do you see
yes there's no no person who made these
captions
yes was machine-made yes it's true we
did it
business by me yet we got it way
yes I am so excited when
so proud players to announce that we are
launching this
this speech recognition and you too
combined
to make captions way and first a little
enjoyment let me add something else here
literally you all and we're very excited
to have you here
are from my perspective it's time
we have spent twenty years
trying to build a programming model
it's the as good
yeah that's good completely machine
generated
speech recognition to English to French
I i mean you think about that when it
means
wonderful and I just love this internet
it's just so exciting to see this
and I've been dreaming about this day's
for so many years than it does so
excited to see it actually happened
I do need to caution you a little bit
for this very early launch
you can only see those
machine generated captions we call auto
automatic
captions either captions you can only
see them
for 13 educational partners
who have agreed to join into this early
launch
of course we want to increase that
number add more channels and partners as
fast as we can
I want to show you how people will turn
on this feature when it becomes
available
may do another one here kills are
located at that URL
we will take submissions for the contest
in August judging will occur to the fall
and will announce our winners at the end
of it all caps is not now you notice a
new option
it says transcribe
audio on the pic that one
were comes up as the box
I'm not sure if all you can read it so
our last interpreter
says transcribe audio is an experimental
service that uses Google's
speech recognition technologies to
provide automated captions for video
their parts will come up
every time I know I might be annoying
but it's very important because
being an early and new launch we want to
make it clear to people
that these captions are not from the
owner other video they have nothing to
do with us
we're trying to generate these for you
asbestos we can and you'll see why
we've learned a few things from American
Idol
and so we will have user votes as part
of the judging and we think this contest
is going to be really really exciting
at number two I'd like to announce that
damn
we're gonna give away to every
conference attendee today
of four thousand now say another ticket
still coming on here this example
this man is talking about
giving away Android phones and is also
giving away
something else that's pretty interesting
we went one step further
I included in every device I
he's giving away salmon
sorry I have not today though really he
is talking about
Sam as I am the same cards that we put
in phones
context you didn't
sorta figure it out but you will see
many of those types of mistakes
now leads me to
another big thing that I wanna talk
about here and that is quality
remember our mission
is to try and make all the information
accessible to everyone and that includes
video
but right now weekend only recognize
English and do this machine translation
for English
and within the English videos some are
not very suitable
because there may be background noise
man includes music
that's noisy strong accents
strange vocabulary so you will see a lot
of
different captions that are
what theyre speaking book we are
concerned about the quality
in fact during the development of this
we had many discussions at Google in you
too
ask yourself this is good now should we
wait until we have improvements
she would launch now I kept saying no
don't worry about it
i like it I love it compared to nothing
this is wonderful it's important to me
that I know for any video I can get
the best quality captions possible I
know they won't be
greater perfect but I know over time we
will get them better and continue to
improve
that's important will continue improving
means fortunately
Google and YouTube have a strong culture
as been already mentioned
we launch early and quickly we do beta
luncheons as fast as we can get it out
there
that's means the faster we get feedback
to people like this do they not like it
what do they like what do they want us
to
focus on what we work on next
so your partner this experiment
in fact that's a big reason why we have
invited all love you here today
so you can help us you're the first
people to see this
and we want you to give us feedback
will have plenty of time for that
during a Q&A I hope and also after that
one last thing I wanna an
about how to improve the quality
you know machine generated captions
these automated captions are not perfect
they will
it takes a very long time before it will
ever reach the quality of human-made
caption
so what are we gonna do about that
so for looking at this I'd like to
take the problem in certain look on the
other side MC
we've tried to lower the bar for
accessibility for those who are watching
the videos
but what about the people who own videos
who are uploading the videos
how do we lower the barriers for them to
help answer that question
one of our that'll be our next feature
speaker
I kinda teased her I'd given her the
title its caption evangelist
and this is Naomi Naomi
are you standing here thank you can
so I start my tiny little bit
about me I'm a ten honors I need him how
many you know what that means
no idea okay how many fewer than ever
uploaded a video TTM so
olive you action honors a 10 honor is
just somebody has logged in to YouTube
and uploaded a video so I'm a channel
owner for Google channels on YouTube
channels
so you can imagine I applaud a lot and
video content
so what I'd like to talk to you about
today
is as a channel on YouTube there are
three things
first I'm gonna show you how captioning
works now in UT
for everyone second I'm gonna show you a
new feature
another new feature also from speech
recognition
that helps channel owners and third
I'm gonna show you how the automatic
captioning can already showed you
how that helps channel owners with their
content caching efforts
so first of all I'm gonna show you how a
channel I know what happened captions
now
this isn't new this has been working on
YouTube since 2008
so if I wanna put captions the first
thing that I need
is captured by now let me show you what
that looks like
signed one thing this file on my desktop
it's called Eric keynote
dot SBB and thus a caption file
there's no one caption format am this is
an example that one but in fact there's
no standard
am but most cash in fast look pretty
similar
mike is on it is on excellent okay I'm
gonna come over here and I'm gonna show
you what's in this package you probably
not familiar with that
at the top of the file I have some
numbers in fact that line contains two
numbers separated by a comma
the first number tells YouTube or any
video player when to start playing the
text
the second number tells it when to stop
playing the text
and then below it we have text in this
example these captions actually tell me
a little bit about what's going on
the first text says man and then there's
a colon so I know who's speaking
and then there's a line break and it
says ladies and gentlemen and there's a
comma
and then the next caption starts and you
can see again we have a time could
and as we go through the file out this
file is just text I can open it in any
text editor
as we go through the file we have every
single caption that I want to show in
the video
so anybody who can prepare if I like
this can I captions to their video
in this much easier than doing video
production
let me show you how this works something
go back and you team here
closes goodbye for now
and I'm going to find that video
a scientist for got much panelists thing
on a kid's parents can't sell it scare
good arm I am
keen it datya
is that she wanted yes I happen to my
channel which is called the Google
developers channel
to queue in to YouTube dot com such
Google developers you would find my
videos
and mitzi your tissues the keynote
you've seen this video already right
new member I'm 10 and
okay so I click over here and captions
and subtitles:
and this option appears because I'm
video honor
and I click adding captions a transcript
and I choose my file and I go over here
here's my final air keynote that is
being
to use it given the language
and member can show you video that had
lots of different languages I can add
many different languages here and so I
tell you what language the videos and
its English they want change then it's
okay
up let it go there it is so here I can
click on English
and here's my time had by all so here we
are in you to you
I have time codes on I've start time
code and timecode here's the text now
YouTube knows all that information that
was in my text about
and it can use it to play this captions
on the video this is my file
I own it so if I want to I can go
download the caption file again later
I don't remember where I started and I
lost it i cant take
that file I can send out to translation
and then I can get a French version
under Spanish version or maybe cling on
Virgin
at a I don't think going on is
officially supported on YouTube
but we could we get up let them am so
let me go back
okay so that's how we uploaded captions
to video on YouTube
and when i saw this feature i got really
excited and this is how I
met Ken and how I became caption
evangelistic Google
I thought this is amazing I have to go
forth and caption every video the Google
post because this is so easy and so
awesome
we have to go do this so that worked out
really well for me
but I looked around and I noticed they
weren't a lot of channels they were
catching their videos
and I wonder why that was and I think
the reason is
uploading videos to YouTube is really
easy it's simple
preparing a time code file is hard for a
lot of people
imagine am a young teenager whose
uploading videos for their friends to
YouTube
they're not gonna have this kind of
expertise and the prime I can be an
interest in learning
so could we make this any easier the
answer is yes
so now I'm gonna show you a second new
feature am this new features called
automatic priming for captions on
YouTube and this is something that
all channel owners once we launch this
later this week all channel owners on
YouTube will be able to do this
question is okay um to do this
rather than a caption file when I'm
gonna show you the transcript
I am here's an example one for a video
that was shot last week
this is a video that we're gonna post to
the Google Blog and it shows you how to
use all these features
so here's the text we prepare the text
ahead of time because we're shooting the
video we didn't wanna make mistakes
and we filmed this text we have somebody
speaking in the voice over
so I already have this it's not gonna go
in to YouTube
gonna get my channel is my videos
and click here on captions am this is
just if you're channeling you might be
familiar with this year this is where
you can see all your videos
an issue the button says captions but
it's actually taking AZT the same page
here
and now I can click adding captions a
transcript
and she's a file just like I did before
and I say okay now I have this option
here
week our old file that we did that was a
caption I'll but now there's another
option YouTube it says transcript
I'll it says english-only because this
is relying on speech to text technology
and right now we can only do that for
English
an essay okay a plan file get 'em
this is gonna take a little outside
gonna talk am
wanna things and i think is important to
mention ken is already told you I
captions
community it's like mine around am ken
is already told you
my captions are important for viewers
and you know whether important for the
risk is your viewers
but I'm a channel owner so I wanna tell
you why this is important for me
cash is important to me because I wanna
reach a bigger audience
if I caption my videos I can reach deaf
and hard-of-hearing
audience I can reach people who may be
immigrants to the United States who read
English well
but don't understand it well when it's
spoken I can reach people
who have found my video except they did
a search and the caption
back ima you can see this video owner by
adding a cash in fact my video
whether I'm doing it from a caption
feiler transcript as I'm about to show
you
adding these captions make that video
much more useful a much more accessible
to really wide audience
sodomy is a channel honor this is really
important and exciting
because I want my videos to be watched
by as many people as possible
okay when you actually go back
actually it's done okay I'm gonna do
this in different order to the finish so
quickly
thing on let me show you its what just
happened
some member we uploaded that video file
let me let me show you the file in case
you forgot my miss
I was talking about how awesome captain
Zack so his with the pilots like great
this is just text
and I go here
and I have this new team you can see
their time cuts here
so what YouTube did is it took my
transcript file and it went through it
and it said
when were these words spoken aloud in
the video and computed
a time code and the calculated the time
code for me
so I don't have to go to all the travel
to figure out where this time codes
aren't
at an abandoned you ever tried to make a
caption about but it's a little bit
fiddly
arm and so YouTube did all this for me I
had to have was the text and I already
had it
so now I have a caption file i can
download it let me show you
good as its gonna my desktop let's go
abandon
hears about its got captions that SBB
that text
out but captured bout
let me show you the side by sides you
really see the difference
okay so here on the left see you guys
are planning means I'm line at here with
the mic
I I really have to thank the guys on the
Google idea team could beat
about the amazing speech to text
technology that makes this happen
and the people at YouTube and people
like and there's a lot a piece is a
Google it all came together to make this
happen
understanding up here with the mic tell
you how awesome this is so
let me tell you how awesome this sewer
here we have the time kurds
you can see if you look at the original
transcript its actually put in line
breaks for me
where I had line breaks its respecting
them you can see today starts on its
online
and again in the caption file it starts
on its online so
if I as the video honor want to control
with line breaks are I can put in line
breaks
and I'll come through the caption and if
I don't know what I'm doing and I don't
care I can make one big one paragraph
and it'll break it into nice link
captions so I can out
I downloaded this far from you to him
you saw that wasn't very hard and I can
take that file
I can send it for translation I might
want to adjust weirdest
caption breaks are I can do that cuz I
have all the time code it's not gonna
show you one last thing
let's go back to supplying let's go back
on the way let me show find that many M
and do not

I and unit I wanna find a video that
that can show you it with the
contentious speaking went scenes 10
two hundred years away the pounds Eunice
okay
so I'm channel honor I own this video
and this video happens to have been
machine transcribed
using and the caption transcript feature
that can show it to you
so if I go into captions and subtitles:
just cooking here's the honor
I see this track here it is English
Machine Transcription
what is this this is the speech to text
transcript
that was provided for me so I can click
on this
you see here I have time codes
I have text no person created these time
coded captions
in some cases they'll be pretty bad in
other cases like this one they're
actually quite good
so I can go over here and click download
let's go back to my desktop define it
when it go I it's this one I think after
this one is BB
Curtis so here's the time coded file
produced by machine and I can go in and
I can find that line about the salmon
I can fix it and i can reupload it so
what this technology has done as 10
liners it save me an awful lot of typing
so I just showed you some new features
from the channel on your side that we
really hope are gonna help
March and honors add more captions their
videos and with that
I'd like to his back to men well
you've just seen how powerful computing
technology can be
its especially when
or with different parts I've
communications
can be linked together and here you seen
the merging
a a various kinds a technology
speech to text we don't have got
text to speech demonstrated here but we
do that to
what's important here is the in
neighboring capability
that tell you just seem I'll be very
interested to
hear from Marco clinton others whose
work
involves producing these things my guess
is that this kind of technology
can make this whole process much more
efficient
and therefore much easier and perhaps
even less
costly to accomplish which means more
video
can be camping now we have an
opportunity
I had to do some Q&A I'm going to go to
that in just a moment
I actually have a bunch of questions
myself the
for from my colleagues so it no one else
ass anything I've got a whole bunch of
them just listening
to what's going on here but this is
simply
a stunning example what happens when you
put
technologists out together with
different kinds of expertise
we have research group I'm part of that
group although I don't claim to
contributed very much in certainly
nothing here except enthusiasm
out but what's important is that the
speech understanding group speech
recognition group
I had to come together with the people
who understood ways in which to do camp
shins
here's something interesting about what
you just saw
the way the captions work in the YouTube
world is different
from the way the captions work in
today's video world
you are aware I'm sure the early late
seventies
when we had 121 captioning and we fought
very hard some others
to put that in even though there were
people saying well there were better
ways to do it and we should wait until
video text was available because it
would have a wider variety of the
symbols that could be expressed in most
ways said the heck with that just get
something up so we can
appreciate video like everyone else does
then no course
are where we've come to high-definition
television by the way she tell you that
youtube is just announced that it can
put up
to him ADP highest resolution high
definition
that you can download and watch from
from YouTube so we're moving ahead with
the rest of the industry
but some love you probably know that
high definition captioning
hasn't worked terribly well its I would
say
erotic the Federal Communications
Commission has a whole
group Assange to work on their problem
and I can tell you that
on the half Google I'm going to be
digging is deep into that as they can
but what's very interesting about what
you're seeing here is that the captions
are being delivered
differently than they are with
traditional television
they are distinct and separate channel
so to speak
not in the sense that naomi was talking
about but in the sense that the source
at the captions is independent over the
video
they're obviously went together by the
time codes but because they are
segregated
we can do all these things to manipulate
the text file
we can translate from one language to
another we can automatically
do this partitioning we can give you the
transcript so you can do editing on it
you don't have to have all the kinds of
equipment
that are needed to decode the captions
or two
encode the captions it's a separate
process
this separation up the two media the
text medium
and the video medium creates
opportunities that would not otherwise
exist
and so when you think about things like
I'm
IPTV you should not be thinking
a just the generating and distributing
he kinda condense combined
a medium but rather one in which the
separation is important
because if the ability to manipulate it
and so are we looking towards the future
this medium the ability to separately
manage
these different streams is absolutely
critical
oddly enough there is something kind of
like that
yeah in the video world where you have
different audio always
you can have different channels with
different audio for Spanish in
English and French and so on but here
we're keeping these media distinct in
many people
well I don't have much more to say in
closing accept a I do want to
acknowledge
some people who made this happen and to
do that
if you hold on to the microphone for a
minute
last again okay
so
80 not everyone is here but a lot of
people are here I'm going did name
the folks who are here and plus a few
that aren't I want them to please stand
up
and accept a are
recognition up the wonderful work that's
been done so can hear in steam
we stand up for a minute and have been
when they owe me and tolliver jewelers
collar
okay air I Greg Milam
this great here okay
at Chris Alberti and me he owned by
Kiani
and use
use I hero to talk to say but he is in
hyeres of in Japan but
we thank him for his work till when
Jonas quick finally
other

you see these t-shirts and
they are are indicative
a Google's determination
to make accessibility apart a
real part up its motto in its objectives
in the world
I hope that we're going to see lots of
other t-shirts with lots of other
indications have accessibility progress
that we make a Google way

0 개의 댓글: