Re: [orca-list] Built in Controls for Espeak Pretty Please with Sugar, cheeries, whatever it takes on top :)

[Thomas Ward <thomasward1978 gmail com>]

Storm,

The shortcomings of Speech Dispatcher are well known to me. For what
its worth I am a software developer and an end user, and I empathize
with your and others frustrations with the current technology. In many
ways I have similar frustrations with Speech Dispatcher but for
slightly differing reasons.

Unlike some on this list I don't exclusively rely upon free software
like Espeak. I have purchased a number of commercial text to speech
systems like Dectalk, Eloquence, and Cepstral, and am just as
disappointed with Speech Dispatcher's lackluster support for them as I
am with its support for Espeak. After all, I might even have more of a
reason to gripe since I paid money for what were supposedly better
quality text to speech systems, but am getting lackluster results
because there is a definite and obvious difference in performance with
the way Orca did with those synthesizers using Gnome Speech and the
performance Orca has with using Speech dispatcher. Its clearly a case
of Speech Dispatcher being a less desirable solution for handling
speech drivers.

The point here is this. So far I have heard a few suggestions of
adding direct support for Espeak in Orca. Basically, writing a Espeak
module that gives Orca direct control over the TTS output. While I see
where people are going with this I see two major problems with it.

First, adding Espeak support to Orca assumes everyone uses Espeak, and
only benefits Espeak users. Anyone who happens to use something like a
commercial speech solution like Cepstral Dianne or a free speech
solution like Festival still get shorted. The developer resolves the
problem with Orca and Espeak, which is fine if one happens to be using
Espeak, but does nothing for the rest of us who may have other
preferences.

Therefore unless we are looking into the option of supporting every
speech system directly by Orca I am not in favor of supporting Espeak
directly. Its a lot of work that only supports a single text to speech
system, and there is a more effective way of fixing the problems with
Espeak without resorting to direct Espeak support. Writing a better
alternative to Speech Dispatcher may be such a solution, and not only
resolve problems with Espeak but might offer improvements for
Eloquence, Festival, Cepstral, etc as well.

Second, adding Espeak support to Orca only helps Orca. This does not
offer a means of offering improved text to speech support to other
applications and games. If we think about Orca and only Orca
accessibility we may in fact be shortchanging ourselves in the long
run by excluding more long term accessibility solutions.

For example, as many of you may know Microsoft Windows offers a speech
API, Sapi, which is available to any and all applications. It is used
by screen readers like Jaws, NVDA, and Window-Eyes for traditional
screen reader speech output, but is by no means limited in scope to
screen readers. Jim Kitchen, a popular author of accessible audio
games, has used Sapi to speak various things in his games. Nextup has
a number of speech enabled applications like TextAloud, WeatherAloud,
and NewsAloud that speaks the weather, news, and various other things
aloud via Sapi voices. Microsoft has a map program that speaks driving
or walking directions aloud as well. My point being that text to
speech support has many applications besides screen readers and we
should not shortchange ourselves by focusing exclusively on Orca, and
ignoring the bigger picture here.

Yes, Speech Dispatcher has problems, and here is an opportunity to
address that issue. Rather than including support directly into Orca I
think what we need to do is replace Speech Dispatcher with a more
stable, more reliable, API for text to speech that does everything
Sapi does for Windows users. Yeah, we can use it in Orca for speech
output, but consider developing something that can be used in talking
weather applications, news applications, e-book readers, whatever.
Plus offer support for as many free and commercial speech engines are
out there so that the user can use what he or she wants to use rather
than limiting ourselves to one and only one text to speech system.

I know, for example, I read a lot of books. Quite a number of them are
e-books. However, when reading books I find Espeak gets on my nerves.
Espeak is alright for day to day tasks like checking mail, writing
code, or something like that but not really for reading a good book.
For that I switch over to one of the Cepstral voices which is more
human sounding, more realistic, and is the reason I was willing to put
a little money out for some higher quality voices for Linux. Therefore
if I took time out of my day to write such a speech system I'd want to
be able to be able to choose which text to speech system I want for
specific situations than exclusively devoting all my effort into one
text to speech system which I tolerate more than like.

So to sum up my thoughts let's consider expanding this support a bit,
not focus so much on including Espeak support into Orca, and see what
we can do to come up with a better alternative to speech Dispatcher
rather than putting all or eggs in one basket with Orca and Espeak.

On 8/27/14, Storm Dragon <stormdragon2976 gmail com> wrote:
> Howdy,
> Nearly every other screen reader has multiple ways to interact with speech.
> NVDA has a built in espeak, and the other nonfree readers have synths as
> well, I think most have a built in Eloquence. Never once, have I heard any
> of those users complain that their screen reader has entirely too many ways
> to talk. Also, as far as I am aware, no one has ever said "Man, I really
> wish my screen reader had to jump through layer after layer of junk to
> speak."
> Speech-dispatcher is ok, for minimal usage, but it crashes with alsa, has
> odd puctuation inconsistancies, and is slow as molasses. The last time
> development of speech-dispatcher even reached a snail's pace was when
> open-speech or whatever was called was made because people weree frustrated
> with the lack of progress made by speech-dispatcher.
> I'm not even asking to replace spd. Let the people who like it use it.
> That's one thing that makes Linux rock, there's usually more than one way to
> do things. for some people, speech-dispatcher may be fine. for me, it's
> falling rather short.
> Also, with a reliable way to deliver speech, speech-dispatcher can drag on
> it its currently abysmally slow progress, and we can have fully working
> espeak. Speech-dispatcher has been around and for a long time now, and there
> still not full support for espeak... No way to use variants, and you have to
> hack it to use freasonably fast espeak.
> Thanks
> Storm