In 1967, the documentary Don’t Look Back began with a now-iconic take of Bob Dylan, sound tracked by his own “Subterranean Homesick Blues”. Standing in an alleyway, Dylan holds a set of cue cards which relate to his lyrics, dropping them as the song progresses. Even if you haven’t seen the clip, you’ve certainly seen one of the dozens of music videos and film sequences which reference it.
Why am I writing about ‘60s counterculture in a blog post on subtitling? Well, I have a theory that those cue cards are a rudimentary precursor of the work we do at Broadcast and Media Services… Not really. Dylan’s influence is huge but that would be frankly ridiculous! The real reason I mention it is because, every time someone asks, “So what do you do for a living?”, I wish I had my own set of explanatory cards. This is what they would say:
– Yes! Subtitling is a job”.
– “Manual input only” – A common misconception is that all subtitling is somehow computerised. After all, automated captions are proliferating online and voice-recognition software has advanced to the point where you and your phone can have a cosy chat. When presented with unfamiliar names, broad accents, broken sentences, simultaneous speakers, intrusive background noise or an intermittent audio signal, however, a computer still says “no”. A human can extrapolate, edit and prioritise, providing the subtitle user with an experience broadly equivalent to the hearing viewer.
– “(Not) on the keys…” – More understandably, it’s often assumed that subtitlers have lightning-fast touch-typing fingers of flame. While stenography continues to play an important role, most subtitling these days utilises powerful voice-recognition software. Each subtitler has a personal voice model, which allows us to continually refine our accuracy and add all the vocabulary we might need for particular output. We also have utterances for all of the ‘non-speech’ elements of subtitles – punctuation, colour changes to indicate different speakers, sound effect and music labels, and a “forced clear”, so as not to intrusively obscure any on-screen action. Software commands called macros do most of this work, spoken phrases which our voice model transliterates into coding instructions rather than words.
– Preparation, preparation, preparation” – To misquote a famous phrase, subtitling is 90% anticipation and 10% adaptation. It doesn’t have to be a huge event like the general election or the World Cup. Every piece of subtitling output requires thorough preparation to ensure the necessary vocabulary and precise terminology is “trained in” to the voice model. We can have access to text in advance, particularly on news broadcasts. We edit this text to a broadcast standard and manually cue it to synchronise with the soundtrack. The transmission program also uses house styles to manipulate the raw text and finesse our output. Particular words and phrases require formatting for certain content, but if they appeared elsewhere, it could be misleading or occasionally, plain unfortunate. A typical day of subtitling might include content as diverse as nature programming and live sport, so a house style ensures that the “whales” appear in the sea when you need them. It also stops them forming a scrum at Cardiff’s Millennium Stadium during the Six Nations.
– Mistakes happen, KO?” – Although the subtitle production process is not generally understood, subtitles themselves are becoming noticeably higher-profile. Broadcasters are legally required to subtitle increasing percentages of their output, and the regulatory scrutiny of subtitle quality has accordingly strengthened in rigour. Some mistakes are unavoidable; a pronunciation slip or the wrong homonym produces not gibberish, but an unintentionally hilarious/cringe-inducing sentence. In this age of instant communication, an eagle eye and a screengrab can splatter an ‘amusing’ error across the web faster than Usain Bolt. Next time you see one of these, have a laugh, by all means – goodness knows we do ourselves – but also consider the context. Subtitled content contains up to 10,000 words an hour, so you’re looking at approximately 0.001% of just 60 minutes’ work, out of hundreds of hours we subtitle every day, 365 days a year.
Chloe Gallagher, Subtitler