[SPEECH-DISPATCHER] Gender-neutral language support
In languages like spanish, the use of gender-neutral language can generate some issues with speech translation from text. The main detected problems are
- Sometimes, words are "de-genderised" by substituting some vowels by the letter "x", which always produce unpronounceable words. For example, the word "nosotros" is usually substituted by "nosotrxs".
- Another issue is the "gender unfolding", that is, any time a "genderised" word appears, it is duplicated, so both genders are represented. This is not a translation problem by itself, but some users have complained about how time-consuming is to listen to translations from "unfolded" texts. As an example of this phenomena, the word "nosotros" is usually substituted by "nosotros y nosotras".
- The last issue is about "de-genderizing" words by substituing some vowels by the letter "e". This generate pronounceable words, however, the words generated are usually unexisting, causing confusion among users. An example of this is the
To sum up, I think it will be a good idea to, somehow, give the option to the user to supress the gender-neutral language (in the languages in which it cause issues to the users), maybe by "pre-processing" the text before executing the translation process.
Do you think this is a good idea?
Edited by Samuel Thibault