Nothing can be done. This is the quality of the recognition engine you use right now. Alternatively, you could use
System.Speech.Recognition
,
http://msdn.microsoft.com/en-us/library/system.speech.recognition.aspx[
^] with
DictationGrammar
. This is simpler, but the qualify will be the same.
The engine provide pretty good quality if all the grammars used at a time and taken together are reasonably small, and all the works used in all of them are distinctly different in pronunciation. This is not a case with
DictationGrammar
, where the number of looks-alike is enormous.
A have read the article in Scientific American claiming that original Nuance Dragon NaturallySpeaking provides really smooth dictation, never tried:
http://en.wikipedia.org/wiki/Dragon_NaturallySpeaking[
^].
There is a number of other products I'm not familiar with. See:
http://en.wikipedia.org/wiki/Speech_recognition[
^],
http://en.wikipedia.org/wiki/Digital_dictation[
^],
http://en.wikipedia.org/wiki/Digital_dictation_software[
^].
—SA