I am blind, but I am not a primary speech synthesis user, I prefer tactile braille. However, I know a number of people which are using their speech synthesizers at rates similar to what you described above.

My theory/experience with this phenomenon is, that a speech synthesizer never makes any errors. When it pronounces a word, it will do so exactly the same way everytime the same word comes up. So the learning effect after a while is a bit higher then when you listen to a human. Humans will always have slight variation in how they pronounce the same word. So, as I understand it, you can "learn" to listen to your speech synthesizer on a fast rate more effectively then you would be able to listen to a fast human speaker.

And yes, I also listen to YouTube talks and audiboosk at about 1.5-2x rate. So I guess 80 bits per second is relatively easily doable for the receiver.