The synthetic voices sound like the speakers are drunk, or maybe hospitalized with a sedative-laden IV drip... I guess that's an improvement over older software that sounded flat out robotic.
I'd suggest you should partner with some native speakers, but I don't know how feasible that is.
For what it's worth, the free program Audacity can slow (or speed up) audio very nicely without changing the pitch or other distortions. You just have to be careful to use the advanced 'Sliding Time Scale/Pitch Shift' Effect (choosing zero pitch change, and the same percent slow down at the start and end). The 'Change Tempo' and 'Change Speed' effects are prone to artifacts so don't use those. (I don't know why they don't use the same algorithm as the sliding scale. effect, but they don't.) There are of course also commercial products (e.g. amazing slow-downer) that have this ability.
It's not quite the same as a natural voice but it's better than synthetic drunks.