Article Lead Image

When voice recognition software backfires

Please excuse the voice-o's.


Janet Kornblum


Posted on Oct 18, 2013   Updated on Jun 1, 2021, 3:54 am CDT

OK, people. I feel like you may need a little bit of a tutorial here. I’ve written about my voice recognition woes before, but I’ve been feeling a need to write about it again. Call it charity.

I’ve been using voice recognition for many years—since the time it was really, really bad, as opposed to now, when it’s only bad sometimes—because I have hands that hurt a lot from typing. It used to be that only geeks and other writers (yes, often the same thing) used the voice recognition software.

So I would have to explain to people, usually via cryptic email signatures, why my emails were sometimes filled with mistakes that I had not caught: “My voice recognition software sometimes says things I don’t mean,” or simply, “I use voice recognition software. Please excuse the voice-o’s.”

There were charming errors such as calling someone a slut—because in one iteration of Dragon (it’s improved dramatically with each version), every time you said the word “select” (which you said a lot because it’s one of the main software commands), the software would not select anything, but rather, it would choose to type the word “slut” in your document. It didn’t care whether the document was an email to your boss or a story for the newspaper.

That’s the thing about voice recognition software. It really just doesn’t care. And it doesn’t speak English, although it can fool you into thinking it does. It says words that it thinks it hears. It doesn’t care that it can’t figure out that your name is not “Jim” but “Janet.” Get it? And you pay the price.

So what do you do? You have to tame it.

Or yeah, you might get fired. Or worse, you could be sending one of those emails to your mom.

Sure, some errors are funny. There’s still a small cadre of people who know me by the name of “Anet” because my voice recognition software, for reasons unknown to me except that maybe it hated me, would cut off letters and words in weird ways. This struck the collective funny bones of my friends (you know who you are) and the name stuck. Some of you actually still call me that name and have no idea where it came from.

But now?

The unwashed masses have entered my particular universe. Thanks to Siri and Google’s voice recognition (doesn’t it have a clever name?), otherwise sane people are spending a lot of time dictating to their phones.

And guess what? They’re making stupid mistakes too. Like writing “their” instead of “they’re,” because voice recognition programs don’t know the difference and, as we established before, they really don’t care.

Maybe they just haven’t learned the tricks of the trade yet. So let me give you a tip: When you dictate stuff into a computer (I’m including phones in this category), mistakes will show up in your copy. And most of them aren’t clever or funny.

Most of the time you won’t call people sluts (or slots, as my computer just typed). And most of the time you won’t tell your suicidal friend that “you need to kill yourself” because it rhymes with the thing you actually said to your computer, which was, “You need to *heal* yourself.”

Instead, you’ll  just be making embarrassing mistakes, like saying “your” instead of “you’re” or inserting letters randomly in your copy. Oh, and you will be saying completely incomprehensible things. Like this line that showed up in an email yesterday that I can only attribute to a “voice-o”:

“Since I do not hear from yesterday after my emails…”


So people, do as I have learned to do: Read your emails before you send them. It’s a lot of work and I don’t always do it. But  it might be worth the three minutes you spend.

Or rather, it would have been.

Right after I finished writing this column, I got a response to an email I had just sent:

“This all sounds very interesting, Janet. I look forward to hearing it from you without having to translate through the voice recognition software ;)”


Janet Kornblum is a writer, journalist, and media trainer based in San Francisco and is probably busy making a voice-o or two.

Photo via Phil Campbell/Flickr

Share this article
*First Published: Oct 18, 2013, 12:30 pm CDT