Alternative handling of non-latin characters for TTS

v4u6h4n
I'm wondering if an alternative approach to handling non-latin characters with the TTS engine would be possible? Currently when encountering a non-latin character it reads <language> letter <number>, for example reading Thai letter 324 for every character in a non-latin word, like in this document.

I would prefer that either the entire word is simply read out, or that instead of reading each letter separately something like Thai letters is used to indicate that multiple characters are present.
  • dstillman Zotero Team
    edited today at 5:03am
    What voice? Thai characters seem to work fine for me with Premium voices. Standard voices don't support multilingual text.
  • v4u6h4n
    Standard Voice 1.

    Premium Voices seem fine; they pronounce the words normally.
  • dstillman Zotero Team
    Right, so that's expected.
  • dstillman Zotero Team
    edited 4 hours ago
    We can look into stripping unpronounceable spans of text for Standard, but mostly we just don't make any claim that multilingual text will be readable.
  • v4u6h4n
    Yeah that's totally reasonable. My complaint is just the tedious reading of each character, I don't have any expectation that it actually read the different alphabets, even though its nice.
Sign In or Register to comment.