Motorola Handwriting Recognizer

[mikelove]

With all of this talk of new platforms recently I've been thinking about how the copyright/licensing end of a potential new Pleco port might work, and our biggest problem at the moment with supporting new platforms is the lack of a Chinese handwriting recognizer; obviously nobody's developed one for iPhone yet, and I'm not sure if a UIQ one would work on a touchscreen S60 without significant modification, so it seems like the Hanwang recognizer we use now may not be available for those new platforms for a while yet.

However, it looks like it may be possible for us to bring back the old Motorola handwriting recognizer we used in Oxford Dict on these new platforms, porting it to them on our own from the original source code. This might also be an interesting possibility on Mac OS, since as far as we know there's no Chinese handwriting input available at all for the latest versions of OS X.

So my question to those of you who've used both recognizers is this: how much of a difference do you feel there really was between them? Any specific things you felt were especially good / bad about the older recognizer? Was it accurate enough to meet most of your needs or do you feel Pleco would be much less useful if we replaced the current recognizer with that older one? And would the (hopefully-small) decrease in accuracy be worth it in exchange for getting a handwriting-equipped version of Pleco on iPhone / S60 Touch / BlackBerry Touch (if one eventually appears) / etc?

 

[curwenx]

Just my 0.02 RMB... I find the hanwang recognizer superior, it recognizes my sloppy laowai semi-cursive handwriting very acurately. With the old recognizer I always had to slow down and try to be nice when writing.

How about decuma? It seems to run on many different platforms (for instance, the opera browser for Nintendo DS uses Decuma for handwritten character input. I believe the kanji sono mama rakibiku dictionary on the DS also uses it.) I would rate decuma as less forgiving than hanwang, but definately better than the motorola.

 

[mikelove]

Thanks for the feedback. We've actually gotten a few e-mails from people that liked that aspect of the old recognizer, since it forced them to be meticulous about their stroke order, but yes, fast and accurate recognition is definitely the main goal here. Decuma's something we could consider, I suppose, but I think unless we persuaded Hanwang to do an iPhone port we'd launch the iPhone version at least with the old recognizer and see where that took us before we considered approaching another company.

 

[Aunty]

Also don't overlook its usefulness when you can't read a character properly, often due to a small bold font that blurs the fine detail together on the paper. I can draw the components of the character that are clear, draw part of the semi-clear components, and put little blobs of scribble in the places that are illegible. Usually that brings up the character that I'm looking for. That would be a type of functionality to be lost, beyond the convenience of writing fast.