Slashdot Mirror
Linux Handwriting Recognition
Dark Paladin writes "Cnn.com is reporting that Communication Intelligence Corp. and Middlesoft is working on Linux based handwriting software that will work in handhelds, Internet appliances, and have GTK and QT hooks. Now we can move from carpal-tunnel to writers cramp. " Now I'm curious, what matters more to people, handwriting or voice recognition?
What is this doing here? It never appeared on the main page even.
But, all-in-all, this is pretty cool. Maybe we'll finally see Palm-style Open Souce handhelds.
And, if you're wondering, an open souce handheld would be great. Why settle for what Palm (or Microsoft) thinks is good for you when you can write your own look and feel?! Just think, you could probably add compatibility for both Palm apps and Wince apps. Now all we need is some good touchscreen support, and we're set! (Or is there already touchscreen support I don't know about?)
Another non-functioning site was "uncertainty.microsoft.com."
The purpose of that site was not known.
The other reason why handwriting is better than voice recognition, despite the possible and immediate use in handhelds, is that we seem to be fairly saturated with voice technologies, but nobody has created a decent OCR package. (Unless, of course, SANE has done it and I haven't noticed.) Getting--not *just OCR*--but handwritintg recognition, is something to be pushed. Hearing about the US post office using stuff for sorting is one thing; it's totally another to have advanced stuff on your own machine.
ultimatly, i think that voice recognition will become commonplace, but that there will always be a place for a good hand written letter. but will anyone be able to tell the difference? Theres a big difference between dictated tone (ie what you say) and what you want to be written. A letter usually sounds different when you read it than when you are spoken to by that person. not only will voice recognition software need to become more accurate... but it'll also need to have some kind of filter to 'formalize' dictations. I think :)
-- Give him Head? Be a Beacon?
-- Give him Head? Be a Beacon? :P)
(If you can't figure out how to E-Mail me, Don't.
Anyway, I think handwriting recognition is definitely appropriate for handheld computer applications, while Voice recognition will evolve into regular applications.
Just out of curiosity, How long will it be before everyone forgets how to write?
-- Give him Head? Be a Beacon?
-- Give him Head? Be a Beacon? :P)
(If you can't figure out how to E-Mail me, Don't.
There are three reasons for handwriting recognition I can see off the top of my head:
While handwriting recognition often gets a lot of gee whiz value, I think I'd find truly resilient voice dictation to be more useful. The reason is that the technology applies to more situations. When one considers how much one communicates verbally (well, those of us that haven't locked ourselves into our cellars with our beloved machines), compared to how much people handwrite - the advantages are clear.
/almost/ every situation where one would wish to write. While the two would certainely be complementary - I think voice stands best alone. For instance, PDAs could be made very effectively using only voice dictation. The reason this hasn't been done is prohibitive hardware costs to wield such processing power in a restricted space / weight / power. This *will* be overcome, though perhaps gradually.
/exceed/ most people's typing speeds - handwriting can NEVER do this. Therefore I would say that many forms of typing will be outdated (though for many applications the additional precision is desirable). I cannot see this happening with handwriting.
Aditionally:
1) A *noise resistant* voice dictation technology would apply to
2) It is readily apparent that in the desktop situation (ie non-portable) a voice technology is FAR *FAR* faster than a handwriting system. As voice improves it will
So from my standpoint it seems that Voice Dictation addresses a true "holy grail" of computing accessibility and ease, whereas handwriting recognition fufills a few important niche markets. This would be a tough choice to make except that voice recognition also performs suitably well in 90% of the aforementioned niches.
As much as I have enjoyed handwriting recognition (I'm a happy newton owner), I would trade it in a second for a really slick and accurate voice interface.
-nullity-
- handwriting is a one-handed activity
freeing your other hand to hold a handheld, the telephone, etc.For a desktop, I actually like having everthing in parallel. Touchscreen, mouse, trackball, stylus, trackpoint nubbin, and glidepad. I've never had them all at once, but I've had many of the combinations and they're all useful at times throughout the day. If my hands are on the keyboard, I don't want to take them off, but if I'm standing, talking to someone, on the phone, or my workspace is cluttered, I like to reach for more convenient or demonstrative devices.
All this technology is pretty cool, but I think it will prove unpractival in time. I especially see talking to the computer as slowing things done. Sure, it's good in telephone systems and the likes where there are no good keyboard, but when it comes to the average desktop computer - it would only be a nuiceance. Even in the telephone appliances it can be boring, here in Sweden the railroad has speach recog. ticket orderings, which gets the most crazy ideas about where one would like to go.
...).
;-) ).
Antoher thing about speaking is that, when the user wants a written report or something about that, she / he would generally produce better material writing than speaking (which gives no time to think about things
Handwriting recognition should be more useful, but, as someone else said here, keyboard recognition is probably much faster (and uses fewer resources - plus the ability to backspace
// Simon Kågström. Enters his text keyboard-wise.
For English-speakers, voice recognition and OCR might seem like neat gimmicks, but they're going to be *vital* to bring information technology into places like rural China, where people are lucky to be literate in their own language, nevermind learning a foreign phonetic alphabet and awkward keyboard input methods.
Check out China's up-and-coming domestic computer maker Legend at http://www.legend-holdings.co m/eng/press_centre6.html... Their basic model includes a keyboard, but more centrally - a writing tablet.
I'm glad to see Linux-based voice and writing recognition efforts. Imagine this - Linux bringing the Internet to 1+ billion more people...
The only thing a handwriting input device buys me is the possibility of a smaller machine. Voice recognition, on the other hand, gives me computer input in contexts where my hands are busy - such as while driving. It also allows for an even smaller machine - since the input interface can be a pinhole rather than a surface at least the size of a Postit note. And I can usually speak faster than I can write.
But while voice recognition may strike me as more enabling than handwriting recognition, it's not an either-or issue. They each add a unique capability. Pick either one, and I'm sure there will be a number of applications where it's a better fit than any other input mode.
Handwriting recognition seems like the ideal input mode, for instance, for paper computers. Imagine a pad of Postits, each running PalmOS. B-)
Also: If they ever come up with a handwriting recognition program that can read my writing better than I can, I've got a lot of old notes around here that I wish I could read...
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
Lets look at the options:
.tif files of every document I wanted to scan. I'd move to completely electronic filing (except for those documents where someone/some government agency would want the original copy of something, of course), if only I could throw a search engine and editing capability around it. I'm a pack rat - I've never deleted a nontrivial personal email. But to try and do that with every important paper doc that crosses my path would be impossible without some better way of organizing/indexing them, and that means OCR.
Handwriting recognition:
Great, so far as it helps with OCR on handwritten pages. Not so great, as the primary interface to a handheld computer. As much as I prefer the Palm Pilot as a PDA, I'd much rather have a tiny, two-finger typeable keyboard than use graffiti (or the onscreen keyboard alternative). Graffiti just doesn't work as quickly for me and feels more awkward. And recognizing "natural" handwriting would be even worse. My handwriting goes from left to right; it doesn't stay within the sensory area on a PDA. The "unnatural" graffiti writing actually helps because it eliminates that instinct to move your hand while writing.
Voice recognition:
Sure, would have been great to dictate homework when I was in the 6th grade and didn't know how to type. But today it would have to get punctuation, spelling of homonyms, etc. exactly the way I wanted it to be marginally better than typing long paragraphs. And as for voice commands? Forget it. It's been like three years since I've had my computer situated out of hearing range of any other computer around, and I don't want to work within some wacky scheme to decide which computer is supposed to be listening to me.
OCR:
Now, this would be useful. I'd love to live a paperless life, but unfortunately I live within a society that doesn't see things my way. It would be nice for my "filing cabinet" to be completely electronic, but today that's not really a good option unless I save
Perhaps natural language may not be efficient for computer control. But that's a separate issue from voice recognition input.
We've been programming and controlling computers using UNnatural, contrived languages for a long time - and directing other machines, domestic animals, and human work teams ditto since long before comptuers - or even recorded history. Think about driving a horse - or a car, military command-and-control, coordinating a crew of sailors, or calling a football play.
Voice recognition technology give us a way to capture, lex, and parse vocal gestures. Whether we try to decode the vocal gestures as a natural or a contrived language is an issue that doesn't arise until the parser level.
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
I was just looking through the documentation for my wife's new Palm (and O'Reilly's book on getting the most out of it). It strikes me that it wouldn't work terribly well for languages other than those supported by iso-8859-1 (latin 1). That is an excellent default, but won't support my I18N work completely.
The net will not be what we demand, but what we make it. Build it well.
Handwriting recognition of Chinese characters would be useful for Chinese users, because most people aren't willing to invest the time to learn one of the many Chinese input methods. It's much easier to write the character out than to press a sequence of keys which represent the component parts of the character...
At this point, voice recognition and speech dicatation have gotten good enough for certain narrow applications. Neither by itself is going to be accurate enough for general use because both technologies need to be targeted to a specific context and vocabulary to work well.
On the other hand, if you combine speech recognition with a system of gestures and written jottings for doing corrections on the fly and for "nudging" the interpretive engine in one direction or another, you can probably increase the speed at which fully-corrected speech or text gets input by several orders of magnitude. Such a system gets rid of the need for stopping dictation/writing to go into an "edit" mode.
Current products show a strange myopia--designs that do handwriting or speech recognition as though users are unable to do both at once, partly an outgrowth of these technologies' origins as accessibility tools. That is, while it's terrific that someone with no mobility can use ViaVoice to fully operate any software other than perhaps a raster imaging package, this approach has made these technologies more tedious and linear than they need to be.
Indeed, such a thing may not even need handwriting recognition to get most of the benefit. I'd love to see what could be done to speech dictation performance with a gesture interface implemented on a pen tablet.
handwriting is a one-handed activity
I can see a large market for hand-writing recognition among certain segments of the online population who use their computers for certain activities that use a certain other hand....
"If one is really a superior person, the fact is likely to leak out without too much assistance" -- John Andrew Holmes
Handwriting recognition could be useful for languages with too many characters to feasibly put on a keyboard (such as Chinese), and for OCR'ing handwritten documents. As an input method for languages using a latin-based alphabet, it's pretty useless, for me at least. I can type much faster than I can write, and I think this is true for pretty much anybody with more than rudimentary typing skills. I'll stick with typing, thanks.
As for voice recognition, perhaps there are some limited uses, but I don't see this as being useful for me or most of the people I know either. I don't want to be talking at my computer - this would require either wanting everybody around to hear what I'm saying or being alone. I don't think it'd speed me up much either. I can type around 100-120 wpm, I doubt I can talk much more than 10% faster than this.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10