[Kde-accessibility] ktts - literature

Mikolaj Machowski mikmach at wp.pl
Wed Dec 29 10:50:55 CET 2004


Dnia środa, 29 grudnia 2004 00:06, Gary Cramblitt napisał:
> OK, I've added an Encoding option to the KTTS Festival Interactive
> plugin. Please give the Polish voice a try and let me know how it is
> working.  Be sure that ISO 8859-2 is set (which should auto default).  I
> don't speak Polish, but it seems to be working.

Thanks, WORKSFORME also.

> It is still speaking punctuation however.  Need to figure out a good
> solution to this.  Ought to be fixed in the voice/synth, I think.  Will
> email festival-talk mailing list for advice.

Yes, final solution of this problem should be on voice level but more
I think, I am more convinced pre-plugins would be very useful:

- some time ago I was a helper for an old lawyer. Still clear mind
  but he had problems with sight. I was reading texts for him with
  every punctuation sign read aloud - nothing could be interpreted
  by me. So, speaking of interpunction could be even useful for
  other voices
- special formatting texts, eg. LaTeX. Voices are broken when reading
  LaTeX tags. Simple preparser could remove them
- 'solving' of shortcuts. You were writing about special SNML(?) tags
  for shortcuts, but they should be recognized before. If in source text
  is pure 'blh', without description something should process it and
  it is not important (for end-user) if processing command will be
  s/blh/blah blah blah/
  or
  s/blh/<tag pronun='blah blah blah'>blh<\/tag>/
  In HTML texts you can expect proper <abbr> tags around shortcuts like
  CIA, FBI which are better known in short forms, in case of eg., AFAIK,
  etc., festival/KTTSD will be lost.

> > > The Command plugin already has an option for encoding.  Did you
> > > choose 8859-2?
> >
> > Yes.
>
> Still need to look into this some more.  When you run KTTSMgr/KTTSD, do
> you have LANG=pl_PL.iso8859-2 ?

No, just pl_PL
:r! /etc/sysconfig/i18n
SYSFONTACM=iso02
LANGUAGE=pl_PL:pl
LC_ADDRESS=pl_PL
LC_COLLATE=pl_PL
LC_NAME=pl_PL
LC_NUMERIC=pl_PL
LC_MEASUREMENT=pl_PL
LC_TIME=pl_PL
LANG=pl_PL
LC_IDENTIFICATION=pl_PL
LC_MESSAGES=pl_PL
LC_CTYPE=pl_PL
LC_TELEPHONE=pl_PL
LC_MONETARY=pl_PL
LC_PAPER=pl_PL
SYSFONT=lat2-sun16

This is standard Mdk9.2 file for Polish users. Don't know how in other
distributions. This is old file, knew versions of Mdk (and other
distros as well) are using internally utf-8. I am not sure what happen
if system encoding will be utf-8. Made some 'dry tests' with setting
$LANG to pl_PL.UTF-8 and then setting up KTTSD but that wasn't real
environment.

> > IMO it is important to automatically recognize page/element language
> > to switch to proper voice. In bko is bug report about that but focused
> > on spell checking but interface for language would be very valuable
> > also for KTTSD.
>
> Its on the TODO list.
>
Another idea - allow for switching voices from applet context menu (like
keyboard applets or KRadio).

m.


-- 
LaTeX + Vim = http://vim-latex.sourceforge.net/
Vim Universal Templates 
	http://www.vim.org/scripts/script.php?script_id=1078
CLEWN - http://clewn.sf.net



More information about the kde-accessibility mailing list