No more excuses for reverting to type

PUBLISHED : Friday, 06 January, 2012, 12:00am
UPDATED : Friday, 06 January, 2012, 12:00am


Are you a tour, or a type? (Sorry, that's not what I meant.) By your talk or a type? (No, not even close.) Are you a talker or a tighter?

Welcome to the world of voice recognition software. I'm speaking into my computer using the new Nuance, Dragon Dictate 2.5 for Mac (HK$1,600). This paragraph is going better than my first sentence, which was meant to read: are you a talker or a typer?

While some people spend all day on the telephone, internet and e-mail, I spend most of my time typing. So much so, in fact, that recently I've been aware of creeping tendinitis in my wrists.

That's bad for a writer. A holiday is a great short-term solution, but what about when you have a deadline for your 1,000 words?

With Dragon loaded into my Mac, I wanted to see how long I could go without touching my keyboard. Can I go a whole week without a single keystroke?

The short answer is no, but Dragon has changed the way I work. It could even extend my career, or at the very least, make writing a pain-free experience. (You'll no longer find a 'tortured artist' in my office.)

I'm halfway through this article, but what you probably haven't noticed is that since that first sentence there have been zero mistakes. I have corrected some errors (the above 'what you probably haven't noticed' was initially transcribed as 'keep Robbie have notice'), but about 95 per cent of the words on this page are as transcribed by Dragon.

Speaking aloud is a very different way of composing a sentence, and the biggest challenges are speed and punctuation. Once Dragon is loaded onto a computer, and a wired headset strapped to the user (though an internal microphone can also be used), it's necessary to go through a short training session. It takes less than five minutes to read aloud to the computer, which gathers all of the information it needs to transcribe a specific voice.

Dragon is not good with '-er' on the ends of words, although it's usually mumbled add-ons that can confuse things, and I discovered early on that the software is at its most effective if you take things slowly. That's an easy habit to form while thinking of what to say next, although voicing each punctuation mark is harder to get used to.

Even an incessant typist such as myself needs to do other things besides simply produce words. During a normal day, I will use the internet, e-mail and iTunes all too often, but can I browse and generally operate these using Dragon? This has been the Achilles' heel of previous incarnations of speech recognition software, and while Dragon makes some advances in terms of navigation, voice-activated web browsing isn't on offer.

But it's still a comprehensive experience. A Dragon icon sits on the desktop and works in every application needing text input; composing e-mail is a hands-free experience (although I couldn't issue the 'send' command without using the mouse), while using live chat on Gmail was utterly voice-activated. 'Open iTunes' proved helpful, and I even managed to call up and begin a playlist of The Beatles.

Just like the much-maligned Siri Personal Assistant voice activation software on the new iPhone 4S (and equally hit-and-miss web search smartphone apps from Google and Bing), if I compare Dragon Dictation to human speech, it doesn't come off very well; but that's not what I'm trying to do. I'm trying to cut my time spent typing. The real test is in comparing this document (pre-edit) with something I hurriedly typed up last week. That page was strewn with errors from my own fingers, with at least one spelling mistake or typographical error on each line; editing it took some time.

Can speech recognition render the keyboard useless? For bad typists, certainly; a week into using Dragon, I've hardly touched my keyboard, and I've made far fewer errors than usual. I still can't help typing some sentences, and speaking with punctuation and commands is strange, but forming habits takes time. A laptop might not be as natural a thing to talk to as a mobile phone, but with speech recognition as accurate as this, even the virtual keyboard's days could be numbered.