voice recongnition?

  • Thread starter Thread starter carlos alejandro perez
  • Start date Start date
C

carlos alejandro perez

Hi folks

Anyone can give me a clue about voice recognition on the PocketPC? We need
about 1000+ recognizable words in our app. I attended to Microsoft Faculty
Summit, and someone there suggested me to use a regular server (that is, a
regular PC) to perform voice processing, sending the "voice-grams" to the
server using XML or another online trick.
Any ideas? do you know if voice recognition will be included in , say , the
next two years on the PPC platform? Is the effort to be spent on this issue
be worthy, or should we wait to MS release of this facility?

Carlos
 
There is already have a XML subset for that called VoiceXML,
I dunno what exactly your app is doing, but you may find someting useful
with VoiceXML
You also need a voice synthesis/engine for ur app.
There is also another type called CCXML for voice apps.

Hope this help

Ken
 
VoiceXml is primarily for voice-only apps over the phone.
It is just now starting to make its way to web pages and voice-enabled web
browsers.
It is not meant for CF apps.

that said, i used SALT (VoiceXml competing standard) with CF here:
http://www.brains-N-brawn.com/speechMulti
a similar concept for CF (not using Speech Server):
http://www.brains-N-brawn.com/freeSpeech
custom speech recognition is here (would not work with 1000 words):
http://www.brains-N-brawn.com/noReco

there is SAPI for CE, but its not really accessible for PPC or SP projects.
there are also some 3rd party vendors for doing reco on the device (i've
never used them);
you should definitely tell them that you want to reco 1000 words.

Thanks,
casey
 
Take a look at Casey's project (below) where he uses FFT "fingerprints" for
words. You could try doing a similar thing - record the word's FFT then
compare against that. I believe that's pretty much how voice dialing on
phones works as well. OpenNETCF.MathEx has a couple FFT implementations
that would get you on the way (and I actually implemented them specifically
because of Casey's work).

-Chris
 
Thanks a lot, great support.. :)
Carlos

Kenneth said:
There is already have a XML subset for that called VoiceXML,
I dunno what exactly your app is doing, but you may find someting useful
with VoiceXML
You also need a voice synthesis/engine for ur app.
There is also another type called CCXML for voice apps.

Hope this help

Ken
Click to expand...
 
Back
Top