Re: [orca-list] Login Desktop, CMU Arctic, SAPI via Wine?

[Michael Whapples <mwhapples aim com>]

Hello,
A lot there, so taking it bit by bit.

First the login issue. The X window manager can be selected via a text
based tool (well it is possible for me using slackware), the tool is
xwmconfig. I have only used this from a text console, but can't think of
a reason why it won't work from a terminal window. If this doesn't work
and you need access to a login window there is information out there on
the internet (I think even at the Orca wiki) describing how to set up
accessible login. I don't know how much of a difference you will notice
if you are using orca in XFCE as it will still be using the GTK
accessibility stuff.

You then went on to TTS. If by the arctic voices you mean the multisyn,
then tese perform very slowly and I would say are very unlikely to
satisfy you for use with a screen reader. I am unsure why it isn't
showing up in gnome-speech.

It may be to do with your configuration, but I find that espeak is
fairly responsive enough, and feel that some of the lag is from Orca,
at-spi and earlier on in the chain. For some the maximum speed of espeak
may not be enough.

Another possibility you may wish to consider (not knowing if finish is
available, although I think it supports quite a number of languages) is
IBMTTS (what was viavoice, and is the same as eloquence). Unfortunately
IBMTTS will cost something, and not being opensource (so you get the
binary only) means that it has some dependencies on particular versions
of libraries. I haven't purchased it myself (as espeak is enough) so I
can't say what the maximum speed is.

As for your final topic of using sapi voices in wine, can't say on that
particular set up. What you could do is have your linux machine and a
windows machine (may be windows in a VM, so long as it can be reached
via a network connection) and you could get speech-dispatcher to send
output to the windows machine (I have a patched version of client.py for
speech-dispatcher which allows customisation of where speech-dispatcher
output should be sent) and then if you are able you could create a perl
script to behave as the speech-dispatcher server (or if you could
compile speech-dispatcher on windows you could create an output module
for speech-dispatcher) to use Orpheus. I don't know whether that would
actually perform fast enough though.

Don't forget, if you felt the command line was better, it is still
available as a full environment and GUIs such as gnome, xfce are not
required to use Linux. I very much believe that is where the real power
of Linux still is.

From
Michael Whapples
On Mon, 2008-05-05 at 14:34 +0300, Veli-Pekka Tätilä wrote:
> Hi, some sort of Orca related indirect questions. If there's a more specific 
> list for generic things like these, feel free to point me to that in stead.
>
> 1. How do I change the login desktop from Gnome to XFCE experimentally, to 
> test the latter's accessibility? I installed the Xubuntu meta package but 
> cannot find a GUI option under login screen for changing the actual desktop. 
> Most graphical instructions talk about a sessions tab or button in the login 
> screen but I need one of magnification, speech or Braille to use that unless 
> you can give me a sequence of hotkey presses doing the trick. I'm still 
> using GDM.
>
> 2. I installed and configured one of the CMU Arctic voices for Festival as 
> detailed here:
>
> http://ubuntuforums.org/showthread.php?t=677277
>
> Festival sees the voice OK and can use it but I don't see the voice at all 
> in Orca under Gnome speech. I've been unable to find any docs on which 
> config file to edit, to get arctic voices showing, either, and have Googled 
> a lot, actually. I am not using speech dispatcher and the speech test 
> program doesn't show my particular arctic voice. All the Festival voices I 
> installed via Synaptic, however, do show OK automagically, as do the Finnish 
> suopuhe voices.
>
> So, any help in getting this voice to work could be great. Though I reckon 
> there's even more latency than in the other Festival voices. Oh well, even 
> eSpeak is too slow at the fastest setting for me and two laggy compared to 
> Meridian One Orpheus I have in Win32. This is a major productivity, 
> intelligibility and comfort drop which makes Linux feel unfairly inferior. 
> Just imagine a font that would be half as slow to read when you count all 
> the greedy speech interrupting and very fast browsing of familiar text like 
> menus. This is very subjective, of course, I've been using various versions 
> of Orpheus for 10 years, but still, as an end user, I feel this may be for 
> me personally one of the biggest remaining hurdles to overcome in LInux 
> accessibility. Other major stuff such as FInnish speech, some music 
> softtware to play with, and color customization of Gnome, have been taken 
> care of over the years, which is just great. I've already installed 
> Audacity, Ardour and Rose Garden but haven't tested any of them yet. They 
> should use GTk2 to my knowledge.
>
> Which brings me to:
>
> 3. Is it possible to run SAPI voices and Windows-based Perl scripts under 
> Wine? if so I just might be able to, some day, write a Perl script to wwhich 
> the generic speech dispatcher interface sends stuff over the network, which 
> then forwards it to my Win32 SAPI version of Orpheus 3 using COM. Not at all 
> sure about any lag, but technically, this should be doable, and something 
> I've wanted for years. I had that back in Debian years ago, when I ran the 
> thing over a serial link and used a Win32 terminal emulator to connect to 
> it. And man using the command line was so much easier, faster and I even 
> felt more confident. But oh well, I'm getting OT again.