1879019 - Emoji / unicode characters are removed upon translation

Status: RESOLVED MOVED

(Firefox :: Translations, enhancement)

Description

[[:sergesanspaille]]

When translating, say, this page https://linuxfr.org/users/serge_ss_paille/journaux/en-passant-par-le-fosdem-avec-mes-sabots from french to english, the musical note in the title 🎵 is just removed. Such emojis carry a meaning that may need translation (but probably not?) and their removal may significantly alter the meaning of the text. In the example above, it's an hint that the title is actually a parody of a traditional french song. Removing the hint makes the title more difficult to understand once translated.

Comment 1

[Marco Castelluccio [:marco]]

https://github.com/mozilla/firefox-translations-training/issues/261 will help with this.

Comment 2

[Greg Tatum [:gregtatum]]

The model needs retraining with our robustness fixes: https://github.com/mozilla/firefox-translations-training/issues/757