Friday, September 6, 2013

THE DRAGON SPEAKS...APPARENTLY RATHER BADLY...

You may have heard that I finished my first draft of TEA IN TRIPOLI last weekend -- 65,000 words of potential, just waiting for my twitchy fingers to create a fabulous second draft.  Oh, but all those words wore out these fingers, and they are attached to exhausted hands on the end of exhausted arms.  Not quite Carpel Tunnel Syndrome but tired, tired, tired. 

Thus, I decided to invest in a computer program called Dragon Naturally Speaking which can hear what you say and turn it into words, bringing relief to bone-weary hands.  My delightful, beloved ex-husband, John, had an early version of it -- Dragon Speak -- to write his own book, though he had little success because it couldn't recognize his voice and that's fundamental.  However, that was a long time ago and Best Buy was selling the new and improved version 12.

Things went awry from the start: the product rang up as $200 when the shelf read $120.  It was incorrectly labeled on the box.  I paid $120 and headed home.  I cleared the decks, prepared to work, but it took forever to install.  I did laundry and came back; hand-watered the yard, came back; clipped nails (mine and the cat's), came back.  In the end, DNS flipped me the bird, computer-style, a big red circle with a black railroad-crossing X saying, "No way, Jose, I ain't installin' on your stinkin' computer."  I uninstalled, started again.  Another laundry load and an hour of answering questions to justify my right to its installation, it reluctantly agreed to come aboard.  I felt disrespected and a bit hurt.

Once registered, I began the tutorial when the program chooses whether or not it will work with you, i.e. if it will provide "voice recognition."  It's like an audition.  This is where John had experienced difficulties, and where folks said I would "fall down."  But I wasn't worried at all.  I do a lot of voice-over work, have hosted, read news articles, even been resident storyteller on a radio show (sadly defunct) on our local NPR station, Sound Sight.  I have studied vocal techniques and taken singing lessons.  I'm an actress and a professional touring storyteller.  Using my voice is my job.  I had absolute confidence that Dragon Naturally Speaking would acknowledge, nay embrace,  my dulcet tones.  

I explained to its database that I was in America speaking British-accented English.  It didn't question whether I was English, Scottish, Irish, or Welsh, which seemed important, like it might present problems down the line, but okay, if they didn't need it, that's fine.  I put on my headphones-with-mouthpiece exactly as set out in the picture.  Ready to roll.

Lesson 1 required me to say a line which DNS would absorb into its being and store in its database, thereby "recognizing" my voice.  I was to say, "Welcome to Lesson 1" which I did, into the mouthpiece exactly as requested, just like I was at a voice-over audition.  "Welcome to Lesson 1!"  It was not accepted; it was not "recognized."  Ten diverse, discouraging attempts later, a sense of rejection was creeping in.  I decided to use my standard American accent.  Nope, didn't like that.  I tried my Asian American accent which I know to be unrecognizable and I was right.  I tried every American dialect I've ever used, from Texas to Mississippi, from Montana to New York (both Gentile and Jew).  I tried Liverpudlian, Cornish, and cockney; did impersonations of James Stewart, James Mason, and Donald Duck.  DNS wasn't having any of it.  By now, I was offended and upset.

Lesson 1 eventually suggested that my English-accented-American vocals weren't good enough to pass the recognition test.  I had failed the audition.  But I was to be given another chance.  DNS suggested I move on to Lesson 2 which, it turns out, is for those with speech impediments and special needs.  I fully appreciate the value of this and am extremely pleased that the people at DNS take seriously the requirement for such a program.  On the other hand, I've never had a speech impediment except for a bit of a "sh" sound on my "s" when I've been reading aloud for a long time, or I'm weary, or I've had a little too much gin.  And my only special needs are a cup of tea first thing and another at 4:00 p.m.  I tried not to be insulted and did as I was told: read from a piece of chosen text (a book by Dave Barry) for five minutes.

The only word that voice recognition could make out was "the."  I chose another text and read for five more minutes and the only discernible word was "the."  And it's now day #4.  DNS has now heard me read for forty-five minutes and still, "the" is all I am apparently capable of saying.  How can I "improve accuracy" when I'm only saying, "the"?  I mean, what is there to improve upon?  It should be finest, most accurate  "the" in the known universe by now!

I'll close this by saying, I dictated an email using my Smartphone this morning and guess what, the text was as clear as spring water and there was only one misspelled word.  And you know what that means, don't you?  DNS -- all $120 worth of it -- has failed its audition and is not welcome on my computer.  DNS needs to work on its own vocals before it deigns to comment upon mine.  To quote Forrest Gump, "That's all I have to say about that."

1 comment:

  1. You may not be having much success with Dragon, but you have successfully given me the giggles - I am truly sorry for your travails! I hope they refund your money...

    Shefali

    ReplyDelete