This is an annoying quirk of English language, but most of TTSs (including Edge) pronounce longer "shhhh" like /ˈɛs ˈeɪtʃ ˈeɪtʃˈeɪtʃ/. This could be demonstrated with simple:
edge-playback -t '"Hey, shhhhh", she whispered.'
Only shh is pronounced correctly (whatever that means). So I suppose that in the book pre-processing (of course, with an optional switch on, like --shhh) all shh+[^a-gi-z] could be replaced with shh.
Just for record: I myself replace all shhh+s with shush in the exported text file, it is also read quite well by TTS.
This is an annoying quirk of English language, but most of TTSs (including Edge) pronounce longer "shhhh" like /ˈɛs ˈeɪtʃ ˈeɪtʃˈeɪtʃ/. This could be demonstrated with simple:
edge-playback -t '"Hey, shhhhh", she whispered.'
Only
shh
is pronounced correctly (whatever that means). So I suppose that in the book pre-processing (of course, with an optional switch on, like--shhh
) allshh+[^a-gi-z]
could be replaced withshh
.Just for record: I myself replace all
shhh+
s withshush
in the exported text file, it is also read quite well by TTS.