Seeing as speech recognition isn't the sharpest, be it cause of the app, the language support or the speaker, it would be a nice addition to be able to modify the recognized text instead of having to repeat yourself infinite times in an attempt to have the speech recognition accurately pick up on what's being said.