v18.4

This version reverts the secure screen copy fix, like in the original Fastfinge project, because eloquence would fail to work on secure screens.

v18.3

This version reinsures the inclusion of the portable copy fix from the previous version and also includes the recent secure screen copy fix from Fastfinge.

v18.2

Two important fixes from the original fastfinge repository in this version:

1. The mandarin chinese voice now fully supports traditional chinese characters (traditional to simplified conversion for now)
2. Eloquence will now work in portable copies of NVDA, even when it's not run on the original computer on which the portable copy was created.

v18.1

This version includes the fix from the fastfinge repository, which removes the douplicate notice about add-on update success.

v17.4

This version changes the upsampling algorithm for an even clearer output.

v18+upsampling

This version is based on the original fastfinge v18 version, introducing the new stable add-on updater! And of course, it keeps the upsampling, with the latest clarity tweaks. The German translation for the new updater strings have been updated as well. A .pot file is also avvailable for other people willing to translate, preferably into the languages Eloquence supports.

v17.3

This version is the first to have the updater check for updates on hozosch/eloquence_64 instead of fastfinge.

v17.2

This Version improves the upsampling by removing some more artifacts using a notch in the 9 kHz region.

v17.1

This Version has all the fixes from fastfinge/eloquence_64 while preserving the upsampling.

v17+upsampling+multilangdicts

This version drops the upsampling to 22.05 kHz for performance reasons, there is no audible quality loss . Also, the Upsampling is now supported with both 32 and 64 bit, like the rest of the add-on.
Improved user dictionary support is also included. Dictionaries are respected across all languages, and it is enough to switch voices to (re)load them. Switching between voices may take a second longer when there is a lot of dictionary data for a certain language, but it's no big deal. Dictionaries are loaded correctly for each language in return.