In the days when voice recognition software was in its infancy, I remember working with someone who’d always done her own typing but had developed RSI. In order to keep her need to type to a minimum, the company supplied her with an early voice recognition package and gave her an office so as to minimise the ambient noise when using it.
And so, she duly shut herself away, installed the software on her computer and then spent two days chatting to it so that it could learn to recognise her. After much repetition and tweaking (she found the whole process fairly embarrassing and rather arduous!), the software was finally set up and ready to use.
The relationship was doomed from the start, however, as within only a few days my colleague developed a nasty cold. Not only did the software no longer recognise anything she said, but it also insisted on trying to include all her sniffs, coughs and sneezes in her letters! Needless to say, she was less than impressed and decided to use the software only for particularly long documents in future. And, strangely, there was never a document long enough.
This was many moons ago and technology has moved on a great deal since my poor colleague’s fated trial of voice recognition software.
But there are apparently still some issues as I discovered when waiting in a queue recently.
I was in my bank, in a long chain of customers which serpentined its way towards the main entrance. The branch in question had obviously realised that having only two cashiers on at a busy time was likely to slow things up somewhat, so it was airing its own television adverts on large screens in order to keep us entertained. The sound was turned off but on-screen subtitles had been added for us to read.
And entertained we most certainly were, mainly because the adverts had been run through voice recognition software, so the subtitles were a random mix of what was really being said and completely misinterpreted words and phrases!
The other key thing is that only two or three adverts were being played; it was basically a very short looped tape, so we saw the same things over and over and over again.
At first, the queue was silent – people were chatting on their mobiles or texting (mostly to explain why there were going to be late!). I suspect I was the only person looking at, and reading, the screens.
But after the third time of seeing “about” translated as “a trout”, I’m afraid I sniggered. To be honest, I’d assumed I’d misread the first time I saw it. By the second time it came round, I was wondering about a few other subtitle oddities. Only at third viewing could I be sure that it really did say “a trout”!
And so I did gurgle slightly. It was just one of those moments you have to hope you were quieter than you thought. But I wasn’t.
Of course, other people in the queue wondered what was funny, saw I was looking at the screens and started to do the same. After only a couple of minutes, I was not alone in trying not to laugh out loud and, shortly after that, texts and mobiles were being ignored as we competed to point out the next mistake!
On the positive side, I have to say that rather dodgy subtitles created by voice recognition software do seem to work a treat as an ice-breaker; there was quite a bit of inter‑customer chatting going on by the time I reached the front of the queue, and there was no impatience at all. Actually, when the staff came round offering to take cheques so’s to save time, they were ignored, everyone was enjoying the unexpected camaraderie caused by the on-screen entertainment and quite happy to continue waiting in line.
On the down side, the cashiers seemed to find engaging people rather difficult – customers gazed over their shoulders, their eyes still glued to the screens in the hope of spotting a new mistake!
I have to admit, as I left the bank half an hour after arriving with a big smile on my face (though I was rather sad to be saying good-bye to my erstwhile queuing colleagues), I did half-wonder if the bank had included the amusing mistakes deliberately.
Sadly, I now know it hadn’t; I was in that same branch only a few days ago and was subjected to the same looped tape of adverts. But they’d obviously realised there were errors in the subtitles and had had them checked; they are now disappointingly correct.
So, if you’d like to keep your customers queuing patiently, use voice recognition software.
But if you want your audio transcribed accurately, come to Fingertips!