I think the text-to-speech synthesis is very high quality and it’s a very nice addition, but a big drawback is the lack of any control over the playback speed/rate.
For those of us used to consuming audio context at 1.5x or more it can seem frustratingly slow. A playback rate control would allow each user to listen at their preferred speed.
It’s also quite helpful being able to adjust based on the density of the content and the speed it can be absorbed. A shopping list for a 5yr old’s birthday party, and an explanation of the Weinberg–Witten theorem can be understood at different rates…