Enter the Dragon

No martial arts hereNo, I haven't succeeded in reanimating Bruce Lee, but I'm a little bit excited just the same. My wayward parcel turned up, finally, and in it was a disk containing Dragon NaturallySpeaking voice-recognition software. I've played with this sort of application in the past, but none of it was really usable as a substitute for the keyboard. But based purely on initial impressions, this one may actually work.

NaturallySpeaking has been around for some time and has been very positively reviewed, so it isn't like this application recently showed up and no one knew about it. Currently in its ninth release, the software continues to enjoy a high level of popularity, and not only within the sphere of hobbyists; the professional-level versions include packages specifically tailored for the medical and legal professions. It's no toy, and reports of 99% accuracy aren't unusual.

In my case, the current version's system requirements exceeded the capabilities of my obsoletei.e. more than a year oldhardware, but the previous version seemed perfectly acceptable, so that's what I ended up with. So far, version 8.0 appears to be happy in my notebook's 1.3 GHz/.5 gig RAM/XP environment; I haven't seen any ominous blue screens or application lockups, or otherwise noticed any problems. Although, legend has it, the current version of NaturallySpeaking offers a few enhancements and improvements over mine, none of this matters if it won't run properly, and new hardware isn't something I'm willing to contemplate right now. Anyway, I installed the software using the custom configuration option, which saved a bit of disk space. For me, the British and Australian English options are unnecessary, as are a couple other things that would have been included had I chosen the default installation.

Then came the grueling part. In order for the system to understand as many spoken words as possible, it's necessary to read it a predefined story so it can get the hang of the user's pronunciation and inflection, as well as the characteristics of the input deviceusually a microphone or headsetthat will be used. In my case, I decided to use a digital voice recorder instead. This avoids being leashed to the computer with a headset cord, or wearing the equivalent of a small microwave oven on my headI still don't like the idea of wireless headsetsand I can take the recorder with me wherever I go. I probably spent 45 minutes reading to the system, and then it spent about the same amount of time working away on its own, constructing the various databases and files associated with my individual user profile. I note in passing that the current version, according to what I've read, doesn't require this training process.

The acid test, of course, was dictating my own words to see if the system would print what I had spoken on the screen, or merely the skewed approximation I had witnessed from other voice-recognition systems in the past. Out of curiosity, I instructed it to guess at the placement of things like periods and commas, instead of verbally inserting them myself during my dictation. I read it the text of a recent monologue, and here's what showed up on the screen.

Say you have some old notebooks gathering dust in a closet somewhere. Maybe those notebooks contain poetry or short stories from the distant past, and maybe no one is ever read that stuff because well, because it's an old notebooks in a closet somewhere in e-mail conversation the other day reminded me that you don't need a traditional web site to collect those old writings for posterity. All you need is a blog.

I haven't tried the other blogging platforms so your mileage may vary, but I know it's easy to reconfigure blogger for the sort of application. The daily journal style that characterizes the average blog and defines it really is the limitation so much as a convention doesn't have to be used that way. If you turn off or decide the elements that have come to be associated with blog state and timestamps comments track back links, who created the post. You're left with a nice clean space in which to collect the fruits of your labor. Since chronology is no longer a factor. Things like previous posts lists can go

to the use poetry as an example, you can adjust the dates of the poems formally known as posts to change their order on the page. Individual short stories or whatever you happen to be collecting can be similarly arranged photos or drawings would work the same again assuming blogger blogspot is the platform, everything can easily e-mailed to the blog for publication. If you aren't quite ready to show your creative products to the world, you can always make your blog invisible to the masses while you experiment

while disabling the very features that make a blog a blog may seem misguided. There are some fairly attractive blog templates out there. Blog operation in general is straightforward, and it isn't necessary to pay for the privilege. The standard blog configuration is undeniably useful and appropriate for chronological applications. But the few simple adjustments can result in attractive free, easy to use web site for a variety of purposes beyond journaling or aggregating news and opinion. Options. It's nice to have options.

There you have it. Aside from manually breaking the paragraphs, that's exactly the way it came out. Comparing this to the original post, you can see it buggered up a few words and the punctuation needs work, but this was only the first try. I'm encouraged, because with additional training things will only improve.

With any luck, maybe I can soon say goodbye to aching fingers, and hello to mobile dictation. It remains to be seen whether speaking produces the same results as typing; for me, writing is a decidedly visualnot auditoryendeavor. But from what I've seen so far, this ought to at least reduce the need for typing.


No comments:

Post a Comment