Here are two files I just made after i got a notification for this thread. No problem whatsoever.
Tried it with other voices (Tifa, Enid Sinclair, 2B, Beaux ASMR (EN), Marin Kitagawa (ES), Latte ASMR (KR), and everything was peachy.
Besides the info you provided, what model did you use? Just Eng or Multy?
If anyone is looking for a work around, I've had some request to do pretty hardcore stuff with EL, so, using will's example with the word "daddy". I would generate one voice line like in any of the following lines:
'Daddy..'
...daddy...
--daddy--
"...daddy..."
Once my 'dirty words' are generated i would continue with the rest of the script. I use '...' and '--' to create space between words so i can edit the word and put it wherever i want it.
EDt: You can also bypass any sort of filter by using the speehc to sspeech feature
Hope this helps.