What I actually said to my computer: “…foreign voices.” but what Dragon NaturallySpeaking wrote: “…for invoices.”
What I actually said to my computer while creating an article about B12 supplementation: “colorectal cancer risks.” But what my voice recognition software typed: “co-director cancel risks.” Definitely a surprise side effect of too much B12.
The iPhone’s voice recognition is pretty good, but not flawless. Take this example from one of my friends. What she said to her phone: “When should I be at your place?” but her iPhone typed, “When should I be at your penis?”
Particularly awkward since she didn’t catch the error before hitting send.
Unfortunately, I did not catch this voice recognition error before hitting “send” on the email. What I actually said to my computer: I hope you are well, Chasta. But how Dragon, my voice software, spelled that: “I hope you are well, chested.”
Easily the most inappropriate voice recognition software blunder so far this year! Responding to an e-mail from a client, what I actually said to my computer: Thanks for your thorough response! But how my Dragon software translated it: Thanks for your poor response! And no, I did not catch the error – not until the confused client replied, asking for clarification.
My only salvation was the fact that my client has a good sense of humor and assumed I was making some kind of joke.
This type of error is an impossible thing for voice software to do as well as a human. If I’d not spoken clearly enough, even a human might hear the same thing: “poor” instead of “thorough.” The difference: a human can understand the context of social norms and recognize that they must have misunderstood because surely a business associate would not openly blurt out an accusation like that.
But not Dragon NaturallySpeaking. Poor Dragon.
A good candidate for our predictive text/AutoCorrect category, this one from the Facebook post of a friend:
“I’m seriously not that bad of a speller, but when I typed melancholy I really thought it would figure out that I meant MELANCHOLY and not “melon holy”…. Uh… when have you ever heard someone say, ‘You know, you seem a bit melon holy today.’… Stupidy stupid.”
I never heard of such a thing, but apparently Dragon NaturallySpeaking has. What I actually said: “When the plans firm up…” But how Dragon translated it: “When the plants sperm up.” And that was in an email to a business associate. So glad I caught that before hitting “send.”
Dragon NaturallySpeaking excels not only at voice recognition, but at voice recognition errors. Here’s 2 funny ones from this week. What I actually said: “They’ve got a sense of humor.” But how my Dragon translated it: “They got us into humor.”
How dare they!
And, boy, did it go off base on this next one. What I actually said: A Shakespearean technique. How the voice software translated it: “Each experience acne.”
Must have been a young cast, is all I can say.
I know regular subscribers understand, but, for the rest of you and for the record, these posts are the inaccurate transliterations of my voice software, not my actual words. That said, here’s a doozy:
What I actually said to my computer: Jubal Sackett book (a famous novel by Louis l’Amour), but how my voice software translated it: Jew ball sack at book.
To say the least, Dragon NaturallySpeaking (DNS) is not well versed in the novels of Louis l’Amour. And it is definitely capable of being politically incorrect, and frequently stumbles when trying to understand names. Fortunately, I caught this voice recognition software error before hitting “send” on the email.
Here are three items you can find in most health food stores – herbs that are popular with people on particularly healthy diets. Unfortunately, Dragon didn’t have a clue how to handle these.
For example, what I actually said to my computer is “Astragalus,” but my voice software translated it: “as stragglers.”
Another Dragon NaturallySpeaking fail that occurred as I was vocally typing out a shopping list for a friend: What I actually said to my computer: “Get Stevia and…” but how my voice software translated it: “Get Steffi at hand.”
And one final voice recognition software error with health food: What I actually said to my computer is “Ashwaganda.” But how my voice software translated it: “Ashwood ganja.”
Well, many people do call ganja an herb.
Dragon NaturallySpeaking often has trouble with names – personal or brand names. Starbucks menu items are no exception:
What I actually said to my computer: Frappuccino.
How my voice software translated it: Fred Pacino.
A friend of mine recently attempted to get her Apple iPhone’s Siri to understand a common, simple word: barista. No luck. How Siri’s voice recognition translated it: Paris stuff. A friend of hers tried it herself, and Siri failed again: bar restock.
Clearly, Siri hates coffee.
You know that old adage “Feed a cold, starve a fever.” Or is it “Feed a fever, starve a cold”?
According to Dragon NaturallySpeaking, it’s neither. What it typed when I said the former: Feet are cold, star receiver.
Honest, I was talking about the long-snouted creature that eats ants. Clearly, Dragon NaturallySpeaking didn’t understand:
What I actually said to my computer: ant eater.
How my voice software translated it: aunt eater.
I tried once more, and got one more fail: and eat her.
Two recent voice recognition errors that I wish I’d caught before hitting send:
What I actually said to my computer: dinners and galas
How my voice software translated it: dinners and gay laws
And one more:
What I actually said to my computer: Yay, thanks
How my voice software translated it: gay, thanks
Watch out, people – protect your laps!:
What I actually said to my computer: I can film the lap
How my voice software translated it: I can feel the lap
Google’s voice-to-text service uses voice recognition technology to listen to callers and translate their spoken words to text. But it often misunderstands:
What the caller actually said: Hey, Rick! It’s Richard Zaragosa.
How Google Voice typed it: Hey, arrest! It’s torture third goes you.
And one Google voice fail:
What the caller actually said: Hey Rick, it’s Peter at the Funky Bazaar, sir.
How Google Voice typed it: Hey Ricketts Peter but the fund desires her.
Two recent voice recognition errors, just for kicks:
What I actually said to my computer: he kick-starts
How my voice software translated it: he kicks tarts
Those poor tarts! And what kind of eats would you get here, I wonder:
What I actually said to my computer: Let’s meet them at a restaurant
How my voice software translated it: Let’s meet the meta-restaurant
Better to meet there than eat there, I suppose.
Pronounced tuh-BAH-tuh, a Tabata workout is a high intensity exercise routine that … never mind the details; just don’t ask Dragon NaturallySpeaking to spell it right.
What I actually said to my computer: a Tabata workout
How my voice software translated it: ate about a workout
You’ll never lose weight that way. So I tried again, but Dragon wasn’t game for this workout, coming up with: to bow to work out
I’d give up and switch to a workout that voice recognition software could handle, like weight lifting … but it would probably call it wait lifting.
It’s a tough word, even for humans. But for my voice recognition software, it was a complete fail:
What I actually said to my computer: canonicalization
How my voice software translated it: canonical is Asian
I tried once more but the best it could do: cannot colonization