I did experiment with Dorico 1 in connection with Nuance’s Dragon recognition software. I found it very encouraging overall, although designing a useful command set takes a good deal of time (and preferably some additional scripting); and after that, this approach needs also some effort for maintenance, as a command set may evolve for a while.
I don’t do that anymore, since I do so little basic note input, and a voice recognition workflow is not efficient if you don’t keep at it. But for users that do a lot of note input daily, it certainly is worth a try. It comes with a lot of tinkering in the beginning, though; also, I really recommend a professional recognition software. I tried other ones as well (including the Windows on-board package, which is actually a scaled-down Nuance software, if I remember correctly), and they are impressive, but there is a clearly noticeable gap between them and Dragon Professional, for example.
One last thing: a voice recognition workflow can indeed bring some very unique advantages, but the one thing that struck me by surprise was how much it can actually strain your voice after a while.