@HatkeshiatorTND I'm using sherpatts on android, I don't want a clanker voice speakinng to me but something that sounds like a human being.
Piper on loonicks supports espeak ipa symbols but apparenlty not sherpatts (which uses the same models tho)
Also for whatever reasonn speechd STILL doesn't natively support pipertts which is stupid, and the jugaad they suggest basically involve running a shell script each time, which makes it reload the model mulltiple times, instead of idk like running piper tts server and using an api + a reaper process.
It shouldn't be that hard and somebody should have made that in all these years, it woud even be possible in fucking bash