Aggregating Siri on a centralized server cluster is both good and bad. The good is that it can normally get a good syntax formed by comparing speech patterns and you'll have a pretty good dictation of whatever you recorded. That bad part is, the voice recognition technology with syntax is not that much better than normal voice dictation so localizing voice dictation seems to be a better choice right now. Dragon Dictation will do what you want in terms of voice dictation.