Updates to Github #github


espeak-ng@groups.io Integration <espeak-ng@...>
 

[espeak-ng:master] New Comment on Pull Request #747 Fixing sequences of ? and !
By valdisvi:

@rhdunn may review and merge this pull request. But I think, that instead of patching this very complicated logic in pure C, we should refactor espeak-ng that it uses some external/rewritten text tokenization with configuration files (like patterns for dates, phone numbers, abbreviations etc.). The biggest issue I see for Latvian, Arabic and other languages is, that all tokenization rules in espeak-ng are based only around current cursor/character, but tokenization would be much easier to handle, if it would work on the whole sentence (clause in espeak-ng) with rules written in configuration files with something like regular expressions. See e.g. issues #369, #368, #199, #366.


[espeak-ng:master] New Comment on Pull Request #747 Fixing sequences of ? and !
By valdisvi:

@rhdunn may review and merge this pull request. But I think, that instead of patching this very complicated logic in pure C, we should refactor espeak-ng that it uses some external/rewritten text tokenization with configuration files (like patterns for dates, phone numbers, abbreviations etc.). The biggest issue I see for Latvian, Arabic and other languages is, that espeak-ng is not cleanly handling unicode, and its tokenization rules are based only around current cursor/character. Though tokenization would be much easier to handle, if it would work on the whole sentence (clause in espeak-ng) with rules written in configuration files with something like regular expressions. See e.g. issues #369, #368, #199, #366.

Join espeak-ng@groups.io to automatically receive all group messages.