A longtime copywriter friend was recently diagnosed with a kind of “benign” neurological disorder that results in the shaking of both arms. He’ll be fine and typing isn’t impossible, but he’s using Dragon Naturally Speaking (speech-to-text software) anyway, and finding it useful for some tasks.

microphoneBut not necessarily the “better written” bits.


For decades we were told we’d eventually speak to our computers, but speech-to-text conversion proved more difficult than first imagined, and it seems as if computers are only now coming up with the horsepower to get the job done, though even powerful desktop PCs still translate pretty haltingly.

(Note: Before you Apple fanboys point gleefully at the iPhone’s Siri, keep in mind all those commands are processed on remote servers — your iPhone and iPad aren’t really up to the challenge).

Is speech-to-text ever really going to catch on for writers? Typing is a hassle and a real barrier for some, but is dictating a story an even bigger barrier?

Given a clean slate — no “brain training” either way — is dictating a story/copy/screenplay harder or easier than typing it?

Old Dogs, New Speech

I’ve tried speech-to-text and found it difficult to adjust to, though I’ve written with my fingers since the mid 70s and might qualify for incorrigibility under the “Old Dog” statutes.

In other words, after typing for four decades, a quick dictation trial is almost guaranteed to fail.

Fair enough. Yet when sci-fi novelist Charlie Stross decided to write his next novel by typing the narrative bits but speaking the dialog into a speech-to-text engine, it raised a few questions.

Like, is this stuff finally ready for mainstream use? Or is speech-to-text a technology destined for another generation, who will presumably have wired their brains for speech after spending their cavity-prone years talking to their phones and tablets?

The Spoken Word Isn’t Dialog

First, let’s talk about dialog vs speaking. The distinction is important.

Everyday speech is pretty awful stuff. People react badly when they see themselves recorded on video precisely because writers don’t speak nearly as well as we write (video a bunch of teenagers talking and you’ll learn the true horror of the spoken word.)

Then my writer friend wondered if screenwriters — who rely on dialog for so much of their work — wouldn’t find it easier to adopt speech-to-text than a novelist, who generally writes more narrative.

After thinking about it (and recognizing the fact I know damned little about the mechanics of actual speech), I suggested that movie/TV dialog doesn’t really have that much in common with everyday speech.

It has to feel real, but it’s far tighter and far more story-oriented (a quick listen suggest Aaron Sorkin’s dialog [Sports Night, West Wing, The Social Network, etc] runs about 2x-4x as fast as conversations in real life).

Then we wondered if a great public speaker wouldn’t then make a great writer, but — at least according to essayist Paul Graham — great public speakers are often a lot better at motivating than imparting information.

And it’s clear that motivation is not enough to sustain a story; we expect our entertainment to have conflict, an arc and a resolution.

In other words, this spoken word stuff simply isn’t all that simple.

The Nightmare of Speech

In my post about tablet PCs not being formidable writer’s tools, I outlined some of the limitations of speech-to-text (and I’m adding a few new ones):

  • Poor in public places (background noise, and you’ll irritate people)
  • Small PCs don’t yet have enough horsepower
  • Handles accents, dialects and unknown words poorly
  • Is difficult to edit while you’re writing (for those who write this way)

Nicholas Carr’s more provocative headlines aside, he’s assembled an interesting mass of information about the plasticity of the brain.

In simple terms, our brain reacts to different stimulus the way our muscles do; pullups strengthen your biceps a lot more than your legs, and writing by dictation would eventually make your brain far more adept at… writing by dictation.

I’d suggest the technology is still far from bulletproof, but that we’re fast approaching the moment where speech-to-text technology becomes reasonable.

When that moment arrives, will writers adopt it, or find it a barrier to good prose?

Keep typing, Tom Chandler.