The times I have used speech to text software have been hilarious but hardly accurate.
The dilemma they face is between custom recognition and general recognition. Too custom and you can't use them with a cold or if even slightly excited, too general and thery catch everything said within the mike's hearing.
One of the things about EQ is that it is not a dictation environment. Hopefully there is a certain amount of excitement involved. This would tend to work against speech-to-text.
I dread to think of trying to make sense of text coming to group from someone using this:-
"Ink owning an Oak prawn" might be hard to respond to.