Tuesday, March 2, 2010

I am right to you from my speak fascination sofa wear.

I have a new tool in the war chest: speech recognition software. After years of hearing about it, I finally picked up Dragon NaturallySpeaking, which claims to turn your voice into text three times faster than you can type --with 99% accuracy.

I’ve been testing the program for about a week, and it takes some getting used to. At first, my writing came out like email spam from the deposed prince of Nigeria. But the program learns quickly. It adapts to your speaking style, your unique phrase usage— even your accent. It’s eerily good at it catching stream-of-conscious ramblings. It is awkward, however, when you’re writing something with a lot of punctuation, like dialogue. It just doesn't feel natural to say “open quote, that’s great, exclamation point, close quote, he said, period.”

I’ve read that some professional novelists use this program almost exclusively to write their drafts. One famous writer supposedly spends her entire day laying flat other back, talking into her headset. That's not for me. Not only would I fall asleep instantly, but I would miss what I see as a critical step in the creative process: there's something about the deliberate movement of my hand, the friction caused by that extra step, that forces me to think for just a second longer before putting something on paper, and helps me access a part of my brain that can be a little stubborn.

But the program is quite helpful for second drafts, or what I call draft 1.5. When I’m writing a brand new scene, I like to write freely in longhand -- just let it go and see what happens. Then I type it on my computer, repeating some sentences verbatim, reworking others, and skipping some completely. Speech recognition is great for this, because while it feels cumbersome to type something you’ve already written, it doesn’t feel that way to speak it. Besides, one of the best ways to edit is to read something aloud; it’s much easier to pick out the false notes when you have two senses working together instead of one. And once you learn some basic commands (undo, delete this, add that) you can do extensive editing without ever touching your keyboard.

Where I think this speech recognition thing could really take off is blogging, where perfection isn’t the goal so much as getting your voice and your content out there in real time. Blogs are by nature more conversational. It's about showing your authentic voice. And what could be more authentic than speaking directly to your audience? That said, I wrote this on my trusty keyboard. My wife rolls her eyes every time I speak robotically into my Dragon NaturallySpeaking headset. So I don’t think she’d be thrilled about me doing it at six in the morning.


  1. Thanks for sharing. I've been really considering getting Dragon. My dad said he had a version that I could use but it's pretty old and definitely not as adaptable as the newer versions, so I passed on that. We'll see about buying it, though! Let me know if it continues to work out for you.

  2. Sarah -
    I’ve heard that version 10 is light-years ahead of last release – so yeah, the older version may not be worth using.

  3. BC - your writing never ceases to amaze me!
    Happy Blog Reader,
    Kirsten (Porter) Kaiser

  4. You must do your next post in Dragon speak. That would be so cool.