Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

TTS edge case, asking for guidance #55

Open
ThomasR128 opened this issue Aug 30, 2022 · 3 comments
Open

TTS edge case, asking for guidance #55

ThomasR128 opened this issue Aug 30, 2022 · 3 comments

Comments

@ThomasR128
Copy link

I’m not sure if this is the correct repo to post this to, so please move if appropriate.

Recently I have come across some edge cases where TTS software would leave out certain characters:

  • An 18th-century English text that used the long s ſ where appropriate, which in this case I would like to preserve for purely typographic reasons. Same for German texts originally set in Fraktur if one wished to have them rendered exclusively in such a typeface.
  • Some 19th-century German books where the author made idiosyncratic use of the breve and macron diacritics (not used in German), e.g., ă and ā, to indicate correct pronunciation of long and short vowels in footnotes to the text. Replacing them with the non-diacritic forms would remove essential information and render the footnote obsolete.
  • But also, and these are much more common cases: prime and double prime , the correctly typeset foot/minute and inch/second symbols.

I was testing with MacOS VoiceOver, which is all I have available at the moment, and these letters were simply not spoken, so:

  • beſt became bet
  • thou goeſt became thou goat
  • 100′ became 100.

This of course may turn a sentence into utter nonsense.

  • <abbr title="…"> seems to have a perfect 0% reliability of being spoken, and the first two examples aren’t abbreviations anyway.
  • The OTF font feature “historical forms” could be used for the long s. It is all-or-nothing however, so often typographically incorrect (unless one goes to impractical lengths of CSS wizardry), and of course only supported in very few typefaces.

So, what would be the most practical approach to solve this kind of issue, mainly from the perspective of an affected reader?

@mattgarrish
Copy link
Contributor

Ya, I don't really have any good advice for you on this. TTS has largely been a bust for EPUB - PLS and SSML don't work almost anywhere.

The EPUB working group has a meeting with APA group at TPAC in a couple of weeks where they're going to give an update on the pronunciation task force's work. If anything comes out of that I'll let you know, but I'd be surprised if we aren't years away from working solutions.

@ThomasR128
Copy link
Author

TTS has largely been a bust for EPUB

Oh, the KB is certainly clear enough about that. Still, since screen readers internally map and expand some of the most common abbreviations, one wonders why some graphemes of major alphabets are not considered worth mapping…

the pronunciation task force's work

Thanks for pointing to these resources. Good to know that this is in progress.

@mattgarrish
Copy link
Contributor

For VoiceOver, you could always try contacting Apple and letting them know about the problem -- just send an email to [email protected].

I don't know what sort of priority it'd get, but it might be faster than waiting on TTS support.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants