Voice Recognition Software For Journalists

by martinhesp


It was going to be so good – it was going to save me hundreds of hours a year – but alas it didn’t quite happen. For years I have pondered voice recognition software and at last I’ve taken the plunge. The good news is that it does work. For me. The bad news is that it doesn’t work when it comes to other people’s voices.
And, really, that is what I wanted it for. The bane of a newspaper journalist’s life is transcribing interviews. I long ago gave up my old Pitman shorthand favouring various lightweight recording machines instead. The modern digital recorders are amazing, especially for an old radio news hack like me who used to drag around a UHER reel-to-reel machine that weighed as much as a chest of drawers.
The digital recorder I use (pictured) has a hairy windjammer sock over the microphone which means I can interview people even in stormy conditions. I find it much more comfortable simply to chat with someone without the rather unnatural barrier of a pencil and notebook coming in between us.
When I get the interview back to my office I play it through VLC sound software, slowing the speed down slightly so that I can easily keep abreast with the words while typing the interview.
The trouble is, of course, that this takes hours. Laborious non-productive hours. So, imagine being able to let the computer take the strain instead. Feed in the file and tell it to transcribe… What an idea! What a dream come true!
Except it’s not true. Because voice recognition software is just that – it is all about recognition. I have taught my new Dragon Dictate software to recognise my voice – and this it does with rapid accuracy, typing up almost everything I say flawlessly. But put in someone else’s voice – as in the following excerpt taken from an interview I’ve just done with a stonemason – and this is what comes out…
“And down every Avenue it was from there running a now to hear that the candidate as outsiders Samantha van like as go right for the Soo do take on the occasional training or friends is now available on the low last year we tickle a bursary scheme and try a trainee and summative music journalism level to confirm’s NVQ and symmetry and lung he passes level III and 240 we can keep on your say.”Image
Somehow I don’t think my editor at the Western Morning News would appreciate this version of events. Alas, it is back to the long-winded drawing board – or keyboard, to be more precise. But one day soon, surely, there will be voice recognition software that simply picks up everything and anything anyone says.