I submitted a request for audio of the instructional text back in 2015. It is mentioned from time to time with nothing mentioned about planned implementation.
An alternative that I suggested is a “text” flow that would look more like a chat screen. Current text would feed at the bottom, with prior text moving up with each new instruction. I suggested this as it would be far easier to implement via TR App changes since it would simply reuse the existing text already available.
Recording, syncing and applying audio to the workouts (maybe 1000 or so with text?) would be a MASSIVE undertaking.