Had anyone inquired as late as last year whether there were any worthwhile computer programmes that convert speech into text, the answer would have been that that kind of technology was given up for dead years ago.
I type like a chimpanzee. This is due partly to laziness, partly to lack of dexterity - and mostly to a matter of male honour. So I tried throughout the 90s to find some kind of computer utility that would allow me to lie poolside and dictate great thoughts to my laptop while it typed them out flawlessly. I failed.
You still see speech to text programmes in corners of computer stores, but they've tended to end up in bargain bins. The only person I know who's tried one in recent years (because he was suffering from RSI and couldn't type) got nowhere with them.
However, that's all started to change. The processing speed of even the most basic laptop is now dozens of times what it would have been a decade ago, and companies such as Dragon, which has always dominated in an albeit tussocky and weed-strewn field, have in recent months taken advantage of the amazing speed of today's consumer computers.
With some substantial rejigging, speech to text has begun — at long last — to deliver. Indeed, speech to text is suddenly rather hot. Only this last Spring, YouTube introduced an option to automatically convert English speech on video clips they host into text subtitles for the deaf. When you think of the muffled voice quality on most YouTube material, they must be pretty confident that this technology finally rocks.
Still sceptical about the idea, I've been testing the latest speech to text programme from the Dragon stable, coined, significantly, for Macs rather than PCs and being stocked by the Apple stores — an indication if ever there was one that the technology is back and putting up a fight.
MacSpeech Dictate, which is largely the same as Dragon NaturallySpeaking for PCs, has recently come out in the UK. And I have to say, it is an eye-opener, let alone a mouth-opener. Indeed, I think it's going to revolutionise the way I (and I suspect a good few other ape-style typists) work.
After no more than half an hour of training the software to get it used to my voice (and bear in mind, MacSpeech Dictate would improve exponentially if I did more) I have just spent three hours talking to my computer — and seeing all of the 9,600 words
I spoke typed straight into Word with perfectly acceptable accuracy.
You can speak slowly, leave long silences, then gabble away (I'm not exaggerating) and still fail to faze the software. It wasn't that the result was perfect, but it was remarkably close to what I wanted. It would have taken me two days to compose and type such a long piece of work, and still it would have been riddled with typos.
After dictating, it took no more than 90 minutes to run through the 9,600 written words and manually correct all the mistakes. I've known of audio-typists who were far less accurate than these new programmes.
Sometimes, the software was just plain spooky. I'd mention place names you wouldn't expect an American product to handle — Twickenham, Barrow-in-Furness — and it wouldn't stumble. Mikhail Gorbachev, Mao Zedong, Vladimir Ilyich Lenin, no problem. It even, when I got bored and was just playing, managed antidisestablishmentarianism.
Of course, a lot of people will say it's great to know you can finally get speech-to-text programmes that work — but that they couldn't compose their written work on the hoof. It is, indeed, an alien skill.
But if you get properly to grips with the software (ie, read the instructions, heaven forfend), you can do all the corrections you like by voice. It's a new world, but you do need to go to some effort to embrace it. (OK, am I speaking or typing this? Fair question; I'm typing it because I'm working in the premium economy section — and very nice it is too — of a BA flight to Delhi, and I don't expect my fellow passengers would appreciate me dictating as they enjoy the in-flight entertainment.
I have little doubt, having used this new generation of software, that the way we access and create information through PCs and laptops is going to change radically over the coming decade - and that voice is very soon going to become more important than text input.
MacSpeech Dictate and Dragon Naturally Speaking are available from amazon.co.uk or downloadable from macspeech.com
blog comments powered by
