There isn't a voice input in existence that I'd be interested in using while writing code or doing photo/video editing.
Even dictating code to another human who is as fluent as you are in the language of choice is a non-starter, so I don't see any machine based way of doing it catching up soon.
The only solutions I've seen that even "come close" in the speed / accuracy comparison are unfair comparisons.
An example was posted by someone in another comment of a guy doing Python coding with Emacs using Dragon Dictation. In that example, the guy used over 2,000 custom "macros" to get to something close to his pre-RSI speeds. This sounds fantastic, but the speeds of a typer who has learned to leverage 2,000 custom macros would be an order of magnitude faster (imho) than someone dictating using similar macros.
Even dictating code to another human who is as fluent as you are in the language of choice is a non-starter, so I don't see any machine based way of doing it catching up soon.