Does anyone out there have any experiences they’d like to share about sp to txt software programs? We’d like to try some out for a HS student but have heard bad experiences with identifying voice charateristics. There must be something out there that works fairly decent? Any ideas would be greatly appreciated.
and...
… do you have other writing assistive technology? Things like TextHelp! or WriteOutLoud, Inspiration, etc? These can make a huge difference.
Thoughts on Dragon Naturally Speaking
I thought that voice recognition might provide some independence for our then 11-year old dysgraphic/dyslexic son so I bought Dragon Naturally Speaking - Standard Edition.
He was extremely frustrated by it and now only uses it when I insist. And then, for him, it’s really only good for assignments where he is generating content that doesn’t have to be formatted well and doesn’t require appropriate punctuation. (That basically knocks it down to his daily reading reports for the Resource teacher. All the other teacher’s expectations are higher.)
- You have to be highly motivated to do it cause it’s not as simple as just talking off the top of your head.
- I liked Sue’s comment about being highly verbal. The introverts at my house aren’t fond of a lot of talking, especially talking in front of others to the computer.
- You have to enunciate well. My son also has speech/language issues. Combine them with homework avoidance and Dragon avoidance and you get some pretty sloppy enunciation.
- You have to overtly deal with punctuation by saying “comma” or “period.” This does not come naturally, especially if you are my son, who forgets to use punctuation and capitalization anyway. And not using periods in Dragon is worse than not using them in writing. Dragon never knows to initial cap the first word in a sentence if you don’t say “period.” At least my son gets it right part of the time.
- You have to read in many excerpts of text before you can start. That’s pretty hard to do if you’re dyslexic, and Dragon only shows the text in a ridiculously small font size! How stupid.
- It needs a fairly powerful processor and lots of RAM to operate at a reasonable speed. We had to upgrade quite a bit before it would function at a tolerable speed.
- We had to go directly into Dragon after a system start-up (otherwise too many resources were tied up) and we couldn’t use it with WORD (too slow/inaccurate apparently to the amount of system resources WORD uses). Had to use WordPad or Notepad.
- You should use the “Correct That” correction feature to help it fix what it did wrong so that it can be smarter the next time. But it can really be a pain to do it. Our son just gave up.
Our other son (14) enjoyed reading the excerpts to it, got it to work with fairly accurate recognition, and was even good at talking punctuation. But it was amusing for 2-3 assignments and then he just went back to physically writing his assignments.
I got it to work very successfully—I thought it recognized a woman’s voice just fine—but I don’t use it because I hate talking out loud.
Hope that helps.
Re: speech to text
I use a copy of Dragon NaturallySpeaking V5.0 Professional and have used it with students with CP, SB and general LD…(with varying amounts of success). I have found that with this new version it has very good support for a range of speech profiles including younger children. The difference, it seems, between earlier versions and the current is extreme, with the new version hosting a much more efficient interpretation (contextual) engine. Of the other available similar software available on the market ie IBM Via Voice, Dragon works far better when dealing with a diverse range of vocal patterns.
However, if using in an environment where students are either:
- surrounded by peers, or
- being observed in any manner
I have found that their level of accuracy is reduced. To enhance accuracy when using VRS, I have found that so long as the student is comfortable and in a quiet/private area that this is certainly a far greater solution for students who either have difficulties with fine and gross motor control or just generally poor typing accuracy and ability.
I”ve been investigating the same thing for our college computer lab.
I also just got my most recent LDA research quarterly and it included an article about somebody trying voice recognition with four students — for 3 of the 4 it was a complete bust. HOwever, a huge issue was these were middle schoolers. Basically, the stuff is designed to hear adult, male voices. So if this kiddo has that kind of voice there’s a better chance. But it’s important to note that all four students approached it with enthusiasm, adn all four were at best disappointed in the realities and some completely frustrated and not likely to try anything like it in the future.
THere are also, however, persistent technical challenges. And it is *more* demanding on language skills than writing because there is so much mental composition involved; as one student said, “It’s not just talking to the computer.”
PUt “spk2wrt” in a search engine (altavista.com being my standard) and you’ll get a good website about it w/ a discussion archive.
The people who really, really like voice rec software are people who are very verbal but the getting-it-on-paper process is hindered by poor typing or motor skills.
I think Richard Wanderman may have some insights about it, too, on his www.ldresources.com site.