Re: [orca-list] State of accessibility in GNOME/Others, from an absentee

[Thomas Ward <thomasward1978 gmail com>]

Hi Jacob,

No, unfortunately, Speech Dispatcher does not have a native speech
driver for Cepstral which is why the voices aren't very responsive
with Orca. All we have is the generic ci== configuration file which
works for basic speech output, but developing a native driver for
Speech Dispatcher would be nice.

That said, you don't need to pay $300 for the Cepstral SDK and API
documentation. If you look in /opt/Swift/include all the header files
are right there with comments, and the libraries are in
/opt/Swift/lib. All you need to do is add the include and lib
directories to your compilers path and write a Speech Dispatcher
driver that will compile against the headers and libs for Swift 5 or
6. I've looked at doing it myself, and it really does not look very
hard to do from what I've read.

As far as the AT&T voices I've pretty much reached that conclusion
myself. They haven't updated the Windows voices in ages, and I'm
pretty sure their Linux voices are as bad if not worse compatibility
wise.

Cepstral really seems to be the only developer continuing to develop
and support their voices, and who were among the first to switch to
64-bit Windows and Linux while Eloquence, Dectalk, and so on get
further and further out of date every day. So other than ESpeak
Cepstral really is the only other viable choice in my opinion.

On 6/21/13, Jacob Schmude <j schmude gmail com> wrote:
> Hi Thomas
> I'm using Unity-3d on 13.04, and intend to remain there. It's quirky, but it
> works and 13.04 has some updated drivers that I'll require should I decide
> to install it natively. I didn't know about KDE almost being usable, and
> will most certainly check that out. I've always wanted to try KDE in any
> case. Should be fun.
> I wouldn't hold out much hope on AT&T voices. They haven't even kept their
> Windows versions updated, so I'd say for all practical purposes you could
> consider them abandoned. Do we have a Cepstral native speech dispatcher
> driver? All I see so far is the generic configuration file which will give
> basic speech output but, because of a lack of indexing and callbacks to
> communicate with the engine, will fail when doing a say all or other large
> texts. Writing a driver, if we don't have one, isn't that hard and I'd be
> willing to do it, however Cepstral wants $300 for their sdk and API
> documentation.