Speech recognition?

[Elizabeth Dalton]

Hi all,

Thanks to everyone who responded to my questions about using a Director 
Xtra. Guess I'll need to download a copy of gcc (not on my Mac due to an 
unexpected upgrade to Panther during a service call) and see how rusty 
my C skills really are.

Now I have a different question. Has anyone had experience with 
cross-platform speech recognition (not tts)? I have an unusual 
application in mind which may not be able to use commercial products 
like IBM ViaVoice.

I'm working on a set of learning games to help children learn Chinese as 
a second language. I have a number of games designed which help learners 
practice listening and reading, but learners really need to also 
practice speaking (and writing, but that will come later). However, 
unlike usual speech recognition situations, where we want the recognizer 
to try to overlook speech mistakes and optimistically interpret what the 
user is saying, in this circumstance we want to notice common speech 
errors and give the user helpful feedback specific to the error. In 
particular, I want to be able to alert the learners to mistakes in tone 
(a feature of Mandarin Chinese not present in English), vowels, and a 
few key consonants which are common sources of error in beginning 
Chinese students.

Has anyone had any experience with doing sound comparisons or analysis 
directly in RR, or in an external? Can anyone recommend a package or 
library for this purpose, preferably cross-platform and hopefully 
open-source or other non-royalty? I've done some online searches and I 
think I see such packages, but I'd like some hints if there's anyone out 
there who's done something like this already.

Thanks,

Elizabeth