Saturday, November 6, 2010

Google Captions vs. Steno Captions

Yesterday I discovered a wonderful captioning tool by Universal Subtitles (plug).  It enables users to insert subtitles or captions into a video.  I pulled a video from YouTube of President Obama thanking his supporters after the 11/2/10 Election. 

I fired up my captioning software, took a wild guess on how to output to Universal Subtitles captioning tool, didn't read any directions or view any instructional videos (sorry, that's how I roll), and BAM on my first try I successfully captioned the nearly 3-minute video.  Pretty cool!  So what can I do with it?  I tried to re-upload it back to YouTube so deaf and hard-of-hearing viewers could benefit from my captioning as opposed to Google's automated captioning, but the upload failed.  I'm still working on it. 

In the meantime, I posted it to our website here Captioned Videos.  I've decided to periodically donate a captioned video and share it with the public. 

The most interesting thing is I could turn on Google's automated captions and compare them to mine.  So what were some of the errors Google made?  Here are the ones I caught (correct words in parens) [my sarcastic remarks in hard brackets]:

two three (two, three)  [Bascially, there is no punctuation throughout the video, not even a period!]
some of it (some of you)
moment (movement)
believe (believed)
our grandchildren (and our grandchildren)
the President (a President)
white house (White House)
was about (It was about)
in tubes (endures) [my favorite]
was never (It was never)
do it (do.  It)
was not well (was about what we)
engage (engaged)
you knock on doors to make (you knocked on doors and made)
castor votes (cast your votes) [not to be confused with "castor oil."  That'll get the voters out!]
our health report (on health reform)
Wall street were for (Wall Street reform)
I'll still a long road for (on student loan reform)
really positive (real and positive)
eighteen million (80 million) [oops, that's not too far off!!!]
devoted their most precious resource (devoted your most precious resources) [not sure THEY would agree to devoting their most precious resources!]
the energy (your energy)
changes in the state (change isn't easy)
want hard-fought (won hard-fought)
experience that baxter (experienced setbacks)
you're efforts (your efforts)
for the your voice (or that your voice)
new Hampshire (New Hampshire)
home in treating her cancer (home and treating her cancer)
they're small business owners who can't keep their doors open (there are small business owners who CAN keep their doors open)
merely hundred thousand (nearly 100,000)
we're no longer at war in Iraq (who are no longer at war in Iraq) [um, I think the Iraqis would beg to differ!]
I'm asking you for the moment we began this chart (what I've asked of you from the moment we began this journey) [Oh, if only it was a chart we began!  Life would be so much easier.]
I just did my ability to bring about change but in new York (not just my ability to bring about change but in yours) [Not sure Obama wants to tackle change in New York right now.  He has enough on his plate!]
going to do thank you so much (going to do, with you.  Thank you so much.)

So...was it readable for the most part?  Yes.  Do deaf and hard-of-hearing consumers deserve better?  YES.  We all deserve accurate captions, and to offer anything less when accurate captioning exists is unacceptable!  It's not like a mammogram.  "This is the best screening we have available right now," says my gynecologist.  Sorry to bring that up, but it's relevant.  If speech recognition was the best we had, then okay, but it's not.  Stenocaptioning is the best method we have for captioning.  So let's use it!  But it costs more.  Of course it costs more.  It's amazing!  It's like magic.  But it's not magic.  Magic is unexplainable.  Is that a word?  A lot of hard work and training goes into stenocaptioning.  Years and years and years of training. 

I challenge you to watch this video and compare the captioning.  It took me about 15 minutes to caption this video.  After the first take, I went back and made, I would say, four corrections to the text.  I have an average accuracy rate of 99.5%.  And that's for much more challenging content than the President's speech, which is slow and easy to understand.  I caption sports, entertainment, news, meetings, conferences, college classes, and more, sometimes hours on end.

I will continue to donate captioned videos on our website Coast 2 Coast Captioning and maybe someday figure out how to get them back up on YouTube.  If you want to see more accurate captions on YouTube, let Google know.

Until next time...