Dragon Voice Recognition Computer Software
Learn About The Newest Digital Technology
I need some basic info on how DD works with context-free grammars. (We are using groups as equivalence classes and words in the main vocabulary as grammar rules, i.e., if some classes are <People>, <ActionVerbs>, <Objects>, a rule could look like [<People> <ActionVerbs> <Objects>], which would itself be a word in our main vocabulary, and would be heard in Command Mode with macro SendKeys People_1+ActionVerbs_1+Objects_1.) 1) Is there any Kleene star or wild-card equivalent? 2) If a few (minimal number of) words in a group are not ever said in the ordering specified by a grammar rule, does this greatly reduce recognition? Are probabilities learned or are they constantly uniform for all right-words in the bigram (assuming DD computes bigram probabilities)? I also have a question which is not really related to the above: We are using DD 2.01 for windows '95, and I was wondering if a dsp board would speed things up a lot. Thanks very much, --Kavita Thomas kavita@ai.mit.edu the Intelligent Room Project, Artificial Intelligence Lab, M.I.T.