Like the adage goes, you can’t predict the weather, and it seems neither can you predict the accompanying subtitles to report it.
Some of the biggest subtitle blunders come from weather forecasts. Wince inducing, laugh out loud kind of errors that we dine out on across social media. The mistakes are (mostly) funny and in the UK they certainly cheer up the dullest of weekly outlooks. But for people who rely on subtitles, these constant errors hinder their ability to understand a broadcasted programme. Plus with more people using subtitles to consume content and a media industry that’s trying to improve accessibility and digital inclusion, shouldn’t we try and correct so many mistakes? And why are there so many?
The makings of a weather forecast
In the UK, the weather is so much part of our daily conversation. Whether it’s too hot, torrential rain, four-seasons-in-one-day weather or winter for, like, EVER, we should really know the mechanics of how it is presented.
For those like me who are still a little in the dark, it’s made up of a a number of moving parts. These include the weather graphics, projected (sort of) onto green screens behind the presenter, monitors in front of the presenter for sharing the graphics, so they can see the end result, and a hand held clicker to keep the pace of the programming. And that’s about it. There is no auto cue nor accompanying script and the presenter ad-libs the forecast whilst pointing and clicking, as well as listening to their colleagues in an earpiece.
Trickier than you thought? I reckon!
Accurate weather reporting – im/possible?
So without an approved script, the subtitles have to be live streamed, often by automation bots, which can leave the door wide open for subtitle errors. The presenters aren’t 100% fixed into what they are saying and the bots type what they hear. Although this does depend on how well the language and accent of the presenter chimes with the auto transcription software. Have a look at some of the below to see what I mean:
How to improve live subtitles
1. Our weather may be unpredictable but the words used are often the same. Speech recognition technology would allow you to input key words that reflect the vocabulary used. A particular presenter’s unique surname, a very Welsh sounding town, the name of an incoming storm. You get the drift. Much like the weather, prevention and preparation is key.
2. You can also use voice recognition software so that the unique tones and sounds of the individual presenters are registered in the software to make words and sounds more recognisable and to help avoid errors.
3. If the weather report was presented at a slower pace, we could also offer a live notetaker. Not too far off matching each other in pace, TV weather presenters speak at a rate of about 120 wpm which, if you have raced our typists is about the fastest speeds they can transcribe. Perhaps with the traditional lag on live TV, there is scope to try out our transcribers. They’d certainly improve the subtitles’ accuracy, nail the cultural references and be able to make intelligent, instinctive responses in relation to the presenter’s speech, meaning those who rely on subtitles would have no issues understanding what the presenters are trying to say.
See precisely what some subtitle-reliant viewers are engaging with, below.