People generally think and type much faster than they speak so probably not. Plus, no matter how robust speech to text becomes it still struggles with names and slang.
The average typing WPM isn't comparable to the average typing WPM someone has while working on say, an essay. Just like how the average speaking WPM isn't comparable to the same. Even more when what's being written has to be more proper than spoken language, where grammatical errors, broken sentences or stuttering etc is completely normal and easy to ignore.
We've had fairly robust speech to text for a few years now and the only places its seen real penetration in is basic personal assistant stuff involving short and simple queries, so I find it hard to imagine that anyone thinks that speech can replace keyboards as an input method.
Even with people I know who've tried to switch to using speech to text due to injuries, it's been a nightmare because there are no easy ways to use it for anything that isn't plain format-less dictionary language. For casual communication it sucks because people use a lot of slang, often even slang exclusive to their friend group, and for serious communication it doesn't work because "delete last word" or "cut sentence X and paste here" is objectively not as convenient/fast for say, the average western highschool student as doing the same via keyboard.
Not even considering how unrealistic it is to edit/format text via speech.
As a way to reason about this, look at some code (assuming you're a programmer) and try to think of how you would input it without a keyboard.
The main issue with speech to text is that it is not private, and that it pollutes the environment with noise. How would speech-to-text work for a college student making notes in a lecture for instance?
The only place it works is if you have your own private office (e.g. doctors dictating their notes), or at home.
