In Broadcast - March 2014 - 46

I N T E R N A T I O N A L

WWW.INBROADCAST.COM | ISSUE 29 - MARCH 2014

N E W S

InSTUDIO
Voicewriters - The New Captioners For Live Programming
Screen Systems highlight technological developments in the changing world of Closed Captioning...
By Andrew Lambourne
Business Development Director,
Screen Systems

losed captions are great for
the average viewer and a vital
method of access for deaf and
hard-of-hearing viewers. Everyone
sees the beneﬁt - right? The only
problem is there's a growing
shortage of Stenographers, the
specially trained court reporters
who utilise fast steno keyboards
to make words appear on screen
as they are spoken. So what are
the alternatives?
Well, it turns out that computers
are now so good at recognising
speech - not machine translation
directly from the TV, but from skilled
Voicewriters who train their voices
and voice-recognition software to
produce highly-accurate real-time
captions. Live captions can now be
produced by respeaking them into
next generation captioning software.
Screen, part of SDI Media, has been
an innovator in creating captioning
systems that help get the best out of
this new way of creating text.
You might think that there's
nothing to it. Anyone who can listen
and speak at the same time can sit
at a computer and create captions.
Actually it's a lot more demanding
than that; and the key skill is in
that simple phrase "listen and
speak at the same time. "It takes
intense training to reach that level of
concentration and skill - admittedly
it's easier than training to listen and
use a steno keyboard, but it's still
very demanding. And of course if
you slip up or get tongue-tied, the
viewers are left with inaccurate
captions. So, training to be a live,
real-time captioner takes time and

care. Accuracy is always the ﬁrst
focus, but there's more to that than
meets the eye...
One reason steno has been used
for so long for closed captioning
is that, in skilled hands, it's fast,
accurate and predictable. If you
press the right keys (or "stroke the
right chords", as the professionals
would say) and your dictionary
holds a translation from the
phonetic codes to words, you will
get the right output. Gearing up
to achieve this is a long and hard
process - it can take two years
or more to learn the system and
develop sufﬁcient proﬁciency. But
hourly rates for real-time captioning
have been declining steadily over
the last 20 years. Companies which
used to base staff in expensive
ofﬁce suites may now be asking
them to work from home. Many of
them are independent contractors
working for multiple captioning
companies. Providing additional
specialist training is not always
economically feasible. Having posttransmission reviews is now a luxury.
Working from high-quality video and
audio feeds is replaced by an oldfashioned phone line.
Sooner or later the price squeeze
makes it uneconomic for these
skilled people to make themselves
available for captioning work and
they return to the court reporting
ﬁeld or retire, and so the industry
needs to ﬁnd a new solution. This
is where Voicewriters come in.
Training time is more like three
months than two years, though
as with any high-pressure job,

Screen's specialist Polistream module can render audio
snippets from subtitle text via a Text to Speech engine

Q-Live software supports subtitling for live and fast-turnaround programmes,
blending together live and ofﬂine subtitling techniques in a single package

proﬁciency continues to increase
as experience builds. Rather
than having to learn how to turn
sounds into phonetic keystrokes,
a Voicewriter learns how to speak
clearly and precisely so that the
computer does not miss short words
or misunderstand longer ones. And
rather than having to build a whole
dictionary of more than 100,000
words from scratch, Voicewriters
get one out of the box with the
speech recognition software.
Any technical advance like this
has its ups and downs. Speech
is an analogue signal, whereas
keystrokes are essentially digital
(forgive the pun). Speech input
relies on the computer software
making choices between a number
of possible ways of interpreting a
set of sounds or phonemes, rather
than relying on the human being
to press the correct keys to code
the correct interpretation. But of
course the human "writers" using
steno devices will occasionally
make mistakes and press the wrong
keys, just as the speech signal will
occasionally be misunderstood. To
experienced eyes the differences
can be spotted. Ultimately to the
caption viewer, something won't

make sense. Even worse, there
may be a completely embarrassing
blooper. But hey - is it better to
have captions with occasional
errors rather than no captions at all?
As with Stenographers,
Voicewriters still have to carefully
research upcoming vocabulary to
make sure their "voice models"
are in tip-top condition ready for
a show. They still have to practice
and develop the skills to transcribe
a discussion or debate without
distorting its meaning. They still
have the same commitment to
delivering captions which really
make the difference between
accessibility or not. But in doing
so, they may round down a little
to make sure the words ﬁt in the
time available.
It will be interesting to see how
the market adapts to this new
opportunity. Most folks at home
will not even notice the difference.
Screen prides itself in staying on
the leading edge. What matters is
that the industry embraces these
technological developments in
order to address the needs of
not only price and accuracy, but
also making accessible media as
widespread as possible.

Voicewriters have to carefully research upcoming vocabulary to make sure
their "voice models" are ready for a show

http://www.linkedin.com/pub/dir/?first=Phillip&last=Izzard&search=Go https://twitter.com/inbroadcast http://www.youtube.com/watch?v=qpzEatLsHdE&feature=youtube_gdata_player https://www.facebook.com/InBroadcast http://itunes.apple.com/gb/app/inbroadcast/id468652679?mt=8 https://play.google.com/store/apps/details?id=com.branded.inbroadcast&feature=search_result#?t=W251bGwsMSwyLDEsImNvbS5icmFuZGVkLmluYnJvYWRjYXN0Il0 http://WWW.INBROADCAST.COM http://www.screensystems.tv/live-prep-q-live/ http://www.screensystems.tv/subtitling/subtitle-transmission-systems/

In Broadcast - March 2014

Table of Contents for the Digital Edition of In Broadcast - March 2014