BUY NEW BOOKS

New Foreigner Book!

Intruder

a few hardcovers and pbs available from Closed Circle, signed. Latest: Moonlover and the Fountain of Blood, Jane Fancher short story. Chernevog, part 2 of the Rusalka trilogy co-written by CJ and Jane; and Orion's Children, a tetralogy from Lynn. Plus, coming soon: e-books: Yvgenie, and books from Jane.

Archives

CONVENTION APPEARANCES

At Miscon 2013, around Memorial Day, Missoula MT, At SoonerCon, in OKC, around June 15, also Spokon in Spokane, in July/August, Beyond that, we aren't sure.
April 2014
M T W T F S S
« Mar    
 123456
78910111213
14151617181920
21222324252627
282930  

Trying Dragon speech recognition.

I debated for a long time whether or not I could use Dragon speech recognition. I tried it once back in the 70s or 80s and it was awful. You had to talk like a robot to make it understand you. I also hate their run-on commercials.

But I also know that using your hands as I do 24 seven and having a family history of arthritis, especially rheumatoid arthritis, it would be a really good thing not to stress my hands any worse than I need to. S I decided to bite the bullet now that the software has come down in price and get myself a copy.

I read the reviews. I determined that very likely I would be happier with version 11 and 12, explanation being that version 12 gets too helpful.

So I got it and a real nice USB mike. and indeed it is a vastly improved software from the 1980 version. I managed to screw up the USB mic within about 30 minutes of use. Jane tells me it’ll work just fine if I just reboot my computer. I don’t know what I did.

The software is particularly useful for editing. I can send send the cursor through a patch of text and add and subtract words. Unfortunately it doesn’t understand commands like back up and italicize, or even just italicize. Still, it has some virtues.

I’m not sure I’m ready to let it help me voice control my computer. The problems that could generate are too scary. But a day off from typing is not a bad thing.

When editing I use my hands to position the cursor. I highlight what I want taken out. I insert what I want put in. And I handle the bolds and italics.

You control the mic on off by saying or go to sleep. I might want to change that since I write fiction. Also WordPerfect has helps for straightening out spaces between words and capitalizations at beginnings of sentences that make the program work a lot more efficiently.

Anyway, the first trial is a moderate success. Last time, I trained the program for a week and it still couldn’t work adequately for fiction and wouldn’t let me think ‘story’ because I was too busy trying to communicate with the software. It is important to have a microphone that is intended for voice recognition–fewer mistakes.

It’s not quite like having Robbie the robot at your elbow to understand what you meant rather than what you said, but this sure is an improvement on 1980.

30 comments to Trying Dragon speech recognition.

  • Apf

    I’d love to see what it makes out of Atevi names! Better or worse than some copyeditors…..?

  • Walt

    Or even “Bren” let alone “Ana Ismehanan-min” or “Sikkukkut an Nikktukktin”!

    So, I had this inspiration, and expectedly many people have already thought of what took a Southern Californian so long, a heated mouse:
    http://www.amazon.com/s/?ie=UTF8&keywords=heated+computer+mouse

  • CJ

    Lol.
    Actually it managed Irene and Bren without prompting. But once I installed aiji-ma and Cajeiri in its ‘special’ vocabulary, not a hiccup.

    I’m liking this. I’m shocked. It’s got a few bugs: if you shut the computer lid, it forgets where it is and absolutely cannot find itself until you close down and reopen Dragon, then the word processor, then the file.

    Bug two. The word ‘close’ was in the text. I said it, and Dragon helpfully closed my program, at least saving my file on the way. I think I am going to ‘edit’ that command to ‘Bury that bone, Rover.’ or something of the like.

    But it’s real nice to free your mind to consider the reading, just subbing in words where you like.

    And unlike some copyeditors—it never forgets, never misspells, and you get to see what it’s doing.

  • I’m very curious, because this might help me with both speech-to-text and the other way round, text-to-speech.

    How much do you have to direct it, in order to add formatting (bold, italics, headings versus body paragraphs, etc.)?

    How does it do in reading back to you? I’d love something that would read text better. — If Kindle will read back to you, I haven’t found out how. But if Dragon Speaking would, it might be the start of a bee-yoo-tee-full relationship.

    Also — How is it at fielding English versus foreign languages? What if I want to switch back and forth between English, French, and Spanish, say, or others, if I get better at them?

    Yes, I’m also curious how it does with made-up “alien” words, or other created language. What would I do? Add new words to a custom dictionary?

    For that matter, what about the odd historical or technical vocabulary I might want to use?

    Many thanks for the post!

    • CJ

      Any word you enter as ‘your words’ is remembered forever. I don’t know about vocalization: I didn’t explore that. I also opted for old version 11, rather than 12 which has some whistles and bells that have drawn curses rather than praise. But I was quite shocked to find it instantly useable, and, coupled with a word-processor’s fixes for intial caps and such—brilliant.

  • chondrite

    DH tried it back in the early 90s, when it was a novelty rather than a real verbal recognition program. Buggy as all get-out. We got a verbally activated on/off switch for our inaccessible floor lamp (stands in the corner behind DH’s desk), and it works rather like the old DD software. “Computer, lights at 100%.” “Please restate command.” “Computer, lights at 100%.” “Please restate command.” “Computer, lights at 100%.” “Invalid command.” *cursing, and threats to lunge over the desk at the )(*!! switch*

  • Daze

    I’m sure you all will know that Terry Pratchett has been using Dragon for several years now, since his Alzheimer’s got too bad to type. But what he does also use is a front-end to Dragon called Talking Point – see here – which he says made his life a lot easier. May be worth a look.

  • ready4more

    @bcs Depending on the version of Kindle you have you may or may not have TTS. Text-to-speech (TTS) requires a device with speakers or an audio jack. Next go to the screen where you start listing your books and press the menu button. On my 2nd Generation Kindle (and my previous 1st Generation Kindle) directions are accessed through scrolling down to “Experimental” then pressing the middle button. Actually accessing Text-to-Speech (TTS) is accomplished by holding the Shift key (upward pointing arrow) and then pressing the SYM key to stop or play. I don’t know how the later Kindles work and I’ve heard that the Paperwhite version doesn’t have TTS. Be aware that not all books allow themselves to be read… I was re-reading “Invader” by nand’ CJ and TTS didn’t work, but “Chessmen of Mars” by Edgar Rice Burroughs worked just dandy….

    • CJ

      I tried Microsoft Reader’s TTS on Faery in Shadow and it managed Irish names. Which is pretty darned good. I wouldn’t expect a VoiceRec to recognize them without adding them to the dictionary.

  • ready4more

    CJ, I’d think that once you enter an atevi word or name into the dictionary you’d get consistent spelling throughout the document. This might prevent botches of words like wi’itkitiin throughout the print version of Invader. The latest ebook version of Invader has those spelling errors corrected although I’ve found a couple of minor errors that only matter if you are reading word by word (the infamous double words such as the the or so so). Now if Atigeini and Patinandi get standardized…

  • CJ

    Absolutely. Once entered, it’s there for good. And the tricky not-quite-automatic words of English and common typos are impeccable, which saves brain cells for something more useful…like thinking about the plot.

  • ready4more

    CJ, Can you just give an abbreviation and have it spell it out for you? I can see where being able to say DNA and have it put out deoxyribonucleic acid might be helpful in the Cyteen books, or just being able to say T-a would give you Tabini-aiji.

    • CJ

      If I had to deal with, say, Trishanamarandu-kepta too often I COULD do that. As is, I speak fairly fast. And if it falls behind, it’ll catch up while I draw breath. This is not the old Dragon Speech: this thing is taking advantage of modern processing speed.

      • paul

        Does your computer have a dual/quad-core CPU? It’s rather new IIRC. One could be be working rather flat-out on the VR, while the other does the house-keeping, like presentation.

  • CJ, have you ever tried a text expander? I use one daily in my work and it types 45% of my keystrokes for me, greatly reducing the amount of work my fingers have to do. A really good ergonomic keyboard helps too — I’m absolutely in love with my Kinesis Advantage keyboard. It has completely eliminated my hand fatigue and it increased my already fast typing speed 20% within a couple of months of getting it. I was getting 90-92 wpm regularly, now I clock 108-110. When you add on the expander, I now work at the speed of speech with half the required keystrokes and no hand fatigue.

    Voice recognition has come a long way in the last decade. Most hospital doctors are now using it for dictating their medical records. Where it tends to go wrong (badly!) is with names and numbers, and with sound-alike words or words which when slurred together sound like something else altogether. Overall though, with a clear dictator, the VR systems are amazingly accurate and often better than humans at dealing with someone’s particular accent.

  • CJ

    I used to macro all names, places, etc, back when I used Volkswriter: WP is a bit clunky on that and Word is hopelessly clunky. This system, however, got my accent at about one reading of a passage from Alice in Wonderland, and the only thing we have to sort out is an isolated ‘a,’ ‘there,’ or ‘to.’ In context, it does much, much better.

    • Walt

      You could, for example, write “\skn” for “Sikkukkut an Nikktukktin” and “\sk” for the more usual “Sikkukkut”, and then do a search and replace. The backslash would prevent any unintended changes. But it sounds like you don’t need that solution.

  • CJ

    I’m not having to correct it within the word processor—but if I use it on WordPress or Facebook I do. The word processor has its own grammatical/punctuation/capitalization checks. I wonder if your GP has ‘trained’ it to the specific application.

  • Wolf Lahti

    Finally – a post in the writer’s blog about writing!
    Who’d a thunk it?

  • CJ

    Oh, I make them from time to time. ;) I don’t discuss current works because I don’t want to spoil people’s reading when they do get the books. And theory of writing –I’m not sure there is a theory. I’ve done this since I was 10. I do it better, about more complicated people, than I did at 10. But I ought to. ;)

  • Bryan

    I have Dragon Naturally Speaking myself, but being self-conscious on headsets, and the fact that I all but have to yell to get it to understand me…

  • CJ

    Bryan, you need a new headset. I can watch telly with the sound on (though quietly) and mumble to it and it still gets me very clearly. This is what I’m using: it uses the USB and I did once think I was having problems with it, but if a USB headset gets discombobulated, you often just restart the computer and it’ll wake up and work. It’s also the most comfortable headset I’ve ever used—and I spent my youth in nasty headsets in the university language lab. http://www.amazon.com/gp/product/B003VW41G8/ref=oh_details_o00_s00_i00?ie=UTF8&psc=1

  • NosenDove

    You could always get a Grecian slave for your amanuensis but they have become rather scarce.

Leave a Reply