Sonar GNU/Linux merges with Vinux

[Linux for blind general discussion]

Linux for blind general discussion <blinux-list at redhat.com> writes:

> Sure, let synthesizers handle ASCII text, but give synthesizers the
> 	textual pronunciation of Unicode characters, such as smiling face.

Chris Brannon here.
This works fine if you assume that everyone wants English and the
English names for Unicode characters.  It blows up in the multilingual
case.  If I'm reading some text in Spanish and I come across the "pile
of poo" emoji (Unicode character U+1F4A9), I don't want to hear my
synthesizer try to interpret the English description of it according
to the phonetic rules for Spanish.  I want to hear the Spanish words
corresponding to that character.  This blows up pretty quickly, because
there are so many Unicode characters and so many languages.  It is best
to get this right in the synthesizer, once, where it can be used by
every application that needs to speak.  Then pass along the raw
UTF-8 to the synth as-is.

-- Chris