Hello Willie,
Speech Dispatcher does not do any text preprocessing
(as is clearly written in the documentation) and I'm quite
convinced it is a bug that Orca does so.
Take for example when in the following sentence:
Orca (a screen reader) is one of the accessibility tools
When we substitute the '(' and ')' signs with plain text
words:
Orca left parenthesis a screen reader right parenthesis
is one of the...
The result is not a correct English language sentence and
as
such, the TTS system has no chance of getting intonation
and other text analysis dependent things right.
It can be a useful hack for simple TTS with no own support
for such
things, but for the more advanced synthesizers, this
"functionality"
of upper layers makes it impossible to achieve a good
quality
of the resulting speech.
If such substitutions are done by Orca, it is also
impossible to
announce such graphical symbols by sound icons or pitch
changes instead of english names (again, without breaking
the sentence structure).
> Early on, my phone was already ringing because the
experience was different between DECtalk
> users and FreeTTS users.
I find it quite natural that different synthesizers have
different
voices and different ways how to pronounce such things as
character names and punctuation. I even think it is
unavoidable,
because these names are different in each language as well.
Why would this be a problem?
Also, if Orca wants to controll text processing, then we
might speak
about differences in user experience of how the same text is
being read
differently by Orca, speechd-el or Speakup, while the user
experience would
be consistent on a chosen synthesizer otherwise. So the
"problem" doesn't
actually vanish, it just moves to a different place. But as
I said before,
I do not really see a problem.
Now the real problem is when we have to deal with a
synthesizer that doesn't
support such things as punctuation at all or supports them
in a way
that is not sufficient or is not configurable enough etc.
*and* it is not
possible to fix the synthesizer. In such a case, emulation
would seem
as a good, perhaps temporary, solution. But definitely not
in Orca or
Speech Dispatcher or any place where it affects other
components as well.
> For example, it starts becoming very difficult to
> identify where things are occurring.
On the contrary, these things are easy to identify, because
at
least for Speech Dispatcher, we can see the logfiles for
the communication between the client and Speech Dispatcher,
Speech Dispatcher with the output module and output module
with the synthesizer. This way, it is very easy to see what
comes in and out. I'd really recommend to introduce such
detailed logging into the other AT components as well.
There is other problems that are difficult, mainly the
different
kinds of synchronization issues, audio output and others.
With regards,
Hynek Hanke
_______________________________________________
Orca-list mailing list
Orca-list gnome.org
http
://mail.gnome.org/mailman/listinfo/orca-list
Visit http://live.gnome.org/Orca
for more information on Orca
|
