> > I think that was one reason for the EAGI extension. There is a agi > script to pass audio to Sphinx for speech recognition.
CMU's Sphinx if you carfully design the application to limit the number of words that could be spoken at any one time, works. This means re-loading grammers in real time. The other thing that really helps is to run Sphinx in a mode that returns a kind a "probibility tree" and not just the "most probable" list of words. That way you can scan down the tree and pick the first grametrically correct string of words. We humans use grammer rules and sematic context to dis-ambiguate (is that a word?) speech al the time. So if you force users to issue commands using avery structured grammer. you can do well. You can spend weeks and months reading and studying Spinx it is a very large system but lots of documentation too. > _______________________________________________ > Asterisk-Users mailing list > [EMAIL PROTECTED] > http://lists.digium.com/mailman/listinfo/asterisk-users ===== Chris Albertson Home: 310-376-1029 [EMAIL PROTECTED] Cell: 310-990-7550 Office: 310-336-5189 [EMAIL PROTECTED] KG6OMK __________________________________ Do you Yahoo!? The New Yahoo! Shopping - with improved product search http://shopping.yahoo.com _______________________________________________ Asterisk-Users mailing list [EMAIL PROTECTED] http://lists.digium.com/mailman/listinfo/asterisk-users
