Not really. If speech recognition wasn't so shoddy, it would eventually become yet another way to interact with the user interface.
It would become second nature. It wouldn't replace manual interfaces, but it could complement that interface. But for transcription, it is pretty ridiculous. In the same way that we can read silently faster than we can read aloud, typing silently is faster than voice to text transcription.
What's funny, to me, is that twenty years ago I remember reading that voice recognition was just around the corner.
It's a lot harder nut to crack.