> >>What about any multilingual descriptive string which a blind user might
> >>need to hear?
> >This is true, but seems like a rare case (the multilingual aspect).
> Rare is not an acceptable retort; if it can happen, it will happen, and the
> protocol must deal with it.
So what you and Mark are supposing is a case something like the
- I am getting a stream of say, English text with mixed French words,
"My friend said 'Pas que je sache'."
- I understand both French and English.
- I have both French and English text-to-speech software installed.
- My speech software can't detect when it starts hitting words that are
not in its dictionary.
- I hear the phrase as (assuming the common case where the software
spells words it doesn't know):
"My friend said 'p-a-s q-u-e j-e s-a-c-h-e'"
- What I really wanted was to hear:
"My friend said 'paw kuh juh sosh'"
Is this the kind of scenario you are thinking of?
> >It seems like descriptive text would be a good application for rich text,
> >but if that's not acceptable I guess I don't see why plain Unicode isn't
> >good enough for this.
> Again, what happens if the two languages being used are Chinese and Japanese?
> Some sort of protocol is needed to distinguish at least CJK for rendering
> of text (whether on the screen or in speech or whatever). Tagging at the
Why? What is the scenario that is driving this (as per previous
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:34 EDT