AI speech recognition technology vs. reality!

Sometimes they're not even close.

We use voice recognition to "prep" our audio files for transcription. We carefully edit for errors due to:

-Improper use and expectations of SRT’s capabilities.
-Disregard for standards of style, grammar and readability.
-Errors due to poor recording quality, idioms, slang, idiosyncratic speech.
Many of these errors are subtle and many are perplexing. Sometimes they make you laugh out loud! Here are a few!

VR:  And I think by the time I was ten years old, my brother cut down a set of golf clubs for me that were all wooden, wooden shafted, wooden heads—I mean, not wooden heads, but wooden shafts. So he'd read Greek, gripped the handles of the--grips on the golf clubs. And that introduced me to golf. 

REALITY:  What he actually said was So he'd re-gripped the handles.

As far as I know, that doesn’t require a knowledge of Greek!

VR:  He was Dion.

REALITY: He was dying.

Dion, I presume, was alive. But this guy was not.

VR: They did this from Raymond. They did it from from the gardenia.

REALITY: They did this from Bremen. They did it from Gdynia.
(Oh that Raymond and his gardenias!)

VR: We were talking to our kids. 

REALITY: We were towing targets.
I know there are times when it might feel . . . so. . . tempting!

VR:  That happened Raekwon . . . 

REALITY:  That happened right when . . . 
Okay, Raekwon? Really?

VR:  We had to getrillionreports . . . 

REALITY:  We had to you know, get reports . . .

I guess a WHOLE LOT of reports?

VR:  No, I write books! 

REALITY: No, I drive a bus.

Oh, God, I hope not simultaneously!

VR:  Now, we've already done the gag and the garrote.  

REALITY: Now, we've already done the gag (order) and the Garrity (warning)

Sadly, the listener didn’t live to sign anything!