The PPP is not good enough (yet?). Right now, the PPP might pass as a cheap substitute for content creators who need a “voice actor” to read some custom dialogue for quick and free, but for those who already have voice actors/recordings/do their own vocals, the PPP is just a fun proof of concept.

@Background Pony #DDD3  
I wouldn’t say it’s close really, it just does a bit better of a job at splicing a bunch of previous things it has heard together than just raw sentence chopping imo. The tonality is kinda completely off for most things and that can ruin the sort of thing someone is going for in animation. It’s fun for jokes and that sort of thing but if you’re going to actually be putting effort into a thing you’re probably just better off with an actual voice actor even if they don’t sound identical.
 
There are nearly an infinite number of ways to stress and read different words in a sentence and this all depends on what the VA is told to do or the context of the scene they are acting out. The AI has no idea about any of that sort of info so it will never be suitable a human’s skill there unless it is given context or additional instructions somehow on how to go about saying particular words or an overall mood or whatever.

For now PPP is not sufficient enough to create realistic voices, but it’s very close. You can easily recognize a pony by their voice, but there are still two big disadventages:  
>rubber band effect  
>sometimes it’s hard to understand what they say (strange accent, speed)  
Pretty strange to see so many animators saying they will never use PPP. Unless everything what they create is two second animations with SFM futa anthro OCs fucking each other.