I'm curious about how people use speech recognit
I'm curious about how people use speech recognition. I'm just starting out, and my motivations are of course faster text input but also learning to think at the speed of speech, not typing.
Do you dictate an entire document, then go back and fix errors? Or do you go by paragraphs/sentences? Do you use speech to edit, or mouse/keyboard or pen/handwriting recognition?
I'd like to hear all your experiences.
I mostly use it with speech command. It is much more lenient with accuracy. Makes a nice voice-controlled jukebox when coupled with a BT headset and some form of media player. (Set the player to use the soundcard that isn't the headset though).
I have had one rather odd issue with speech command that has happened twice now. It will recognise some word that was the name of a deeply nested folder that I haven't touched in ages or at all!
The most recent one happened around 2 months ago. it recognised the word "Euler", and then opened up a folder "C:\java book\projects\chapter 4\Euler". This was from a set of examples that I had copied from a CD that came with one of my textbooks, and I had copied it over around the start of this year and never opened it!
Andy: I don't use speech recognition a lot, but when I do I try to dictate the entire document (or idea(s) ) at once and only then do I go back and fix errors. I'm one of those people that thinks up the entire composition in one flash, but I can forget it just about as easily so if I don't get it down on paper right away...
Have you found the recognition accurate enough to allow you to go back and figure out what you actually said versus what it recognized? I suppose this approach may require that you have a better idea of what exactly you're going to say. My style of writing is more of a wandering journey; in fact, I usually throw the first draft away and keep the experience of writing it in the spirit of the next draft.
"Have you found the recognition accurate enough to allow you to go back and figure out what you actually said versus what it recognized?"
Yes. When I said that I think of the entire composition at once, I meant that literally. I can dictate entire, near-final-quality documents in one continuous session, and the speech recognition is quite good. However, even when errors creep in, the flow and context is more than adequate to find and correct most errors. Ironically, my biggest problem is that correct or erroneous, everything is spelled right, so spell checking is of only very limited use in tracking down homonyms (words that sound the same but are, in this case, inappropriate...) !! I have to go back and read the text carefully.
Of course, I get a low speech recognition error rate because I have spent a considerable amount of time on the training exercises. I've probably read each of the text selections at least twice, some three times!
Also, just so you don't feel bad, I can FORGET entire compositions nearly as quickly as I think of them, and when I do, I have a LOT of trouble "resurrecting" them, so to speak. I've learned that once something "jells" in my mind, I have only a brief window of time in which to get it written down; otherwise, I forget it again. Thus, a tool like speech recognition is a Godsent for someone like me.
The mind is a fascinating thing, no?
Interesting. How you think is very different from mine. Do you use the built-in speech recognition, or something like Dragon?
Well, for an essay it would have to be sentence/paragraph then using the TIP to correct. I like how voice recognition in MS-Word provides options to alternative words spoken just like the TIP. Easy correction of homonym words as Steve S says.
What sucks is I have to use a headset to input all the words needed for a sentence. It'd be cool if MS-Word could just plug-into my brain and output the words I think then there wouldn't be any spelling errors.
what programs are you using? do you think it would be good enough to dictate a lecture?
If you mean recording lectures and having your tablet recognize what your lecturer is saying, that is very unlikely. You'd have to first get your lecturer to train voice recognition, then ask him to attach a microphone to himself during lectures. Each unlikely by themselves, and even less likely together.
"You'd have to first get your lecturer to train voice recognition, then ask him to attach a microphone to himself during lectures. Each unlikely by themselves, and even less likely together."
...UNLESS, of course, your lecturer was already hooked on TabletPC's...!
krypticide: I use the Windows TabletXP native speech recognition engine. Once trained, I find it to be quite good!
Posting Permissions
Bookmarks