First time dictating a tale
Today, for the first time, I tried dictating a real tale to Dragon NaturallySpeaking. I had already written out a few months ago the material that I was dictating. I wrote it out longhand, and so I had a choice: I could either type this up with my clawlike hands, or I could try dictating it.
This is exactly the kind of thing that I want to use voice recognition software to help me. In the first place, composing longhand accesses a different part of the brain than typing with keys – because when you write longhand, you are actually drawing each letter or word, and this accesses the right half of your brain. Typing or keyboarding, on the other hand, is more digital (Ha-ha ha). Typing seems to use more of the left side of your brain, and so it is isolated from the deep, half conscious right brain.
But then, comes the boring part: once you have your composition in longhand, you still have to get it onto a computer, and typing just copying what you’ve already written is really boring. At best, you can make some changes along the way, but the problem then is that a lot of the time I make changes that I shouldn’t make – because in the very next sentence I find what I thought was missing. And yet, when all you’re doing is transcribing, there is no creativity in what you’re doing. Therefore, the ideal situation is to get it done just as quickly and as efficiently and is easily as possible.
That, I hope means voice-recognition.
Also, it’s always a good idea to read out loud something you’ve written to see how it sounds, to see how the tongue forms the words, and how the ear feels them. In the end, the true tale (and by that I mean the prototypical, original tale) is an oral tale – something to be spoken and to be heard, ideally in the dark by firelight.
So this is what I am thinking, that once I get good with voice recognition software, I’ll be able to transcribe my longhand compositions by speaking them aloud even as I would tell the tale off the top of my head or by memory, and in the meantime I can hear how that sounds and get some feedback on how “true” might tale-telling is.
Already I find that I’m speaking my old writing in a kind of bedside, fairy tale-telling manner – just as I might tell it to a young listener. When I get into that mode I find in fact that the software works better, and recognizes what I’m saying with greater accuracy. I suspect that one of the reasons for this is that I originally trained the program by reading a passage out of Lewis Carroll’s Alice in Wonderland which I read in a similar fashion. I think, in fact, this is probably the way that I would read any tale to any audience. The program is designed to recognize natural speech – that means the natural intonations and phrasings with the rise and fall of the voice. Part of this involves linkages between various words so that the word is not recognized by the software as a discrete tonal unit, but rather as part of a phrase. The final sound of one word will become part of the sound of the next word unless you speak each word with a slight pause in between the syllables – a very unnatural way to speak, and a mode of speech that they say is more likely to strain your voice.
This is – what? – the third or fourth day I’ve been using the program, and already I find that I’m speaking more fluently, especially as I come to trust the program more and more and as the program learns how I talk. Also one of the great parts of this particular software is the way you can feed it a text file of your own writing style, and it will pick up your vocabulary – even the strange names and words that I’ve created. One odd thing that the program did was to start capitalizing a lot of words that I didn’t want to be capitalized. But I think that’s probably just the result of my own strange way of capitalizing some words.
In short, it probably took a lot longer than it should have, because I’m training the program at the same time that I’m using it. And I’m always checking each phrase, each sentence, each line that the program puts down, so there are a lot more pauses in between dictation. In addition, I find the spoken commands to navigate through the text often don’t work even when the program recognizes what I said and will neither type what I said as though it were more dictation nor will it obey the command and treat it as a command. (I haven’t any idea why this happens – then again, I still haven’t read all the manual, and that often helps!)
(Composed Tuesday, September 2, 2008)
