driving a serial synthesizer through speech-dispatcher

driving a serial synthesizer through speech-dispatcher

[Gregory Nowak]

Hi all.

I'm wondering if it's possible to drive a serial synthesizer through
speech-dispatcher? My goal here is to use orca to output to a serial
synthesizer.

While speech-dispatcher itself has no facility a far as I can tell to
communicate through serial ports, something like cat could be used to
send a text of string to the serial port for the synthesizer to
speak. This is crude, and there is no way to interrupt speech until it
completes.

An even better approach is to use
/sys/accessibility/speakup/synth_direct for the speech output. This
would use speakup settings in use for the synthesizer. The problem
with this is the same as with the cat command; no way to interrupt
speech, at least until the todo items in /sys/accessibility/speakup
are resolved, and maybe not even then. Interrupting could probably be
done by sending the silence command for the synthesizer in question to
/sys/accessibility/speakup/synth_direct, to cause the synthesizer to
stop speaking, and flush its buffer. Again, simply invoking speakup's
interrupt command by sending something to
/sys/accessibility/speakup/filename would make this synthesizer
generic.

Has anyone here done what I'm thinking of? If yes, can you please
describe your approach?

Greg


-- 
web site: http://www.gregn.net
gpg public key: http://www.gregn.net/pubkey.asc
skype: gregn1
(authorization required, add me to your contacts list first)
If we haven't been in touch before, e-mail me before adding me to your contacts.

--
Free domains: http://www.eu.org/ or mail dns-manager@EU.org

Re: driving a serial synthesizer through speech-dispatcher

[John Covici]

You might need to write a driver in speech-dispatcher, that would do
exactly what you want.  I have been thinking about this for years, but
never had the time to do it.

On Thu, 06 Feb 2020 19:01:27 -0500,
Gregory Nowak wrote:
> 
> Hi all.
> 
> I'm wondering if it's possible to drive a serial synthesizer through
> speech-dispatcher? My goal here is to use orca to output to a serial
> synthesizer.
> 
> While speech-dispatcher itself has no facility a far as I can tell to
> communicate through serial ports, something like cat could be used to
> send a text of string to the serial port for the synthesizer to
> speak. This is crude, and there is no way to interrupt speech until it
> completes.
> 
> An even better approach is to use
> /sys/accessibility/speakup/synth_direct for the speech output. This
> would use speakup settings in use for the synthesizer. The problem
> with this is the same as with the cat command; no way to interrupt
> speech, at least until the todo items in /sys/accessibility/speakup
> are resolved, and maybe not even then. Interrupting could probably be
> done by sending the silence command for the synthesizer in question to
> /sys/accessibility/speakup/synth_direct, to cause the synthesizer to
> stop speaking, and flush its buffer. Again, simply invoking speakup's
> interrupt command by sending something to
> /sys/accessibility/speakup/filename would make this synthesizer
> generic.
> 
> Has anyone here done what I'm thinking of? If yes, can you please
> describe your approach?
> 
> Greg
> 
> 
> -- 
> web site: http://www.gregn.net
> gpg public key: http://www.gregn.net/pubkey.asc
> skype: gregn1
> (authorization required, add me to your contacts list first)
> If we haven't been in touch before, e-mail me before adding me to your contacts.
> 
> --
> Free domains: http://www.eu.org/ or mail dns-manager@EU.org
> _______________________________________________
> Speakup mailing list
> Speakup@linux-speakup.org
> http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup
> 

-- 
Your life is like a penny.  You're going to lose it.  The question is:
How do
you spend it?

         John Covici wb2una
         covici@ccs.covici.com

Re: driving a serial synthesizer through speech-dispatcher

[Gregory Nowak]

Right, this would require a speech-dispatcher module file. The thing
I'm not sure about is how to do the serial port I/O from
speech-dispatcher. The speech-dispatcher modules are .conf files, so
maybe the answer would be a separate program to expose the serial port
to speech-dispatcher modules, or a modification to the actual
speech-dispatcher code to do that.

What I like about the synth_direct approach is that speakup already
does the heavy lifting, and speakup and gnome-speech would use
speakup's interface to talk to the synth, which would mean they both
wouldn't be setting parameters and causing a mess. The disadvantage is
it requires speakup to be loaded, which I already use anyway, so it
doesn't bother me personally.

Greg


On Thu, Feb 06, 2020 at 11:10:24PM -0500, John Covici wrote:
> You might need to write a driver in speech-dispatcher, that would do
> exactly what you want.  I have been thinking about this for years, but
> never had the time to do it.


-- 
web site: http://www.gregn.net
gpg public key: http://www.gregn.net/pubkey.asc
skype: gregn1
(authorization required, add me to your contacts list first)
If we haven't been in touch before, e-mail me before adding me to your contacts.

--
Free domains: http://www.eu.org/ or mail dns-manager@EU.org

Re: driving a serial synthesizer through speech-dispatcher

[John Covici]

You can treat the serial port like a file (this is linux after all),
just do open in the i nit section and write characters to the file.
The name will be something like /dev/ttyS0 or similar.  Note the
capital S.  As an example look at the espeak  driver, so you get all
the things in the right places.


On Thu, 06 Feb 2020 23:29:43 -0500,
Gregory Nowak wrote:
> 
> Right, this would require a speech-dispatcher module file. The thing
> I'm not sure about is how to do the serial port I/O from
> speech-dispatcher. The speech-dispatcher modules are .conf files, so
> maybe the answer would be a separate program to expose the serial port
> to speech-dispatcher modules, or a modification to the actual
> speech-dispatcher code to do that.
> 
> What I like about the synth_direct approach is that speakup already
> does the heavy lifting, and speakup and gnome-speech would use
> speakup's interface to talk to the synth, which would mean they both
> wouldn't be setting parameters and causing a mess. The disadvantage is
> it requires speakup to be loaded, which I already use anyway, so it
> doesn't bother me personally.
> 
> Greg
> 
> 
> On Thu, Feb 06, 2020 at 11:10:24PM -0500, John Covici wrote:
> > You might need to write a driver in speech-dispatcher, that would do
> > exactly what you want.  I have been thinking about this for years, but
> > never had the time to do it.
> 
> 
> -- 
> web site: http://www.gregn.net
> gpg public key: http://www.gregn.net/pubkey.asc
> skype: gregn1
> (authorization required, add me to your contacts list first)
> If we haven't been in touch before, e-mail me before adding me to your contacts.
> 
> --
> Free domains: http://www.eu.org/ or mail dns-manager@EU.org
> _______________________________________________
> Speakup mailing list
> Speakup@linux-speakup.org
> http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup
> 

-- 
Your life is like a penny.  You're going to lose it.  The question is:
How do
you spend it?

         John Covici wb2una
         covici@ccs.covici.com

Re: driving a serial synthesizer through speech-dispatcher

[Samuel Thibault]

Gregory Nowak, le jeu. 06 févr. 2020 21:29:43 -0700, a ecrit:
> Right, this would require a speech-dispatcher module file. The thing
> I'm not sure about is how to do the serial port I/O from
> speech-dispatcher. The speech-dispatcher modules are .conf files,

Not exactly. speech-dispatcher modules are .c programs, which can use a
.conf file to be configured.

It happens that there is the generic module which can be made to run
whatever command you want by configuring it in a .conf file.

But you can really do whatever you want in an entirely new .c speechd
module.

Samuel

Re: driving a serial synthesizer through speech-dispatcher

[Samuel Thibault]

Gregory Nowak, le jeu. 06 févr. 2020 17:01:27 -0700, a ecrit:
> While speech-dispatcher itself has no facility a far as I can tell to
> communicate through serial ports, something like cat could be used to
> send a text of string to the serial port for the synthesizer to
> speak. This is crude, and there is no way to interrupt speech until it
> completes.
> 
> An even better approach is to use
> /sys/accessibility/speakup/synth_direct for the speech output. This
> would use speakup settings in use for the synthesizer. The problem
> with this is the same as with the cat command; no way to interrupt
> speech,

Indeed. But there is another approach, which was made for this: using
/dev/synth. For now that file only supports writing to it to get
text emitted, but ioctls can be defined to drive it, such as setting
pitch etc. in a portable way across synthesizers (speakup will handle
these details), but also interrupting. That all happens in speakup's
devsynth.c and can be a relatively easy task for kernel hacking
beginners.

Samuel

Re: driving a serial synthesizer through speech-dispatcher

[John Covici]

You could use speechd-up and speakup would not have the device at all,
so you could do what you like with it in your driver.

On Fri, 07 Feb 2020 02:53:46 -0500,
Samuel Thibault wrote:
> 
> Gregory Nowak, le jeu. 06 févr. 2020 21:29:43 -0700, a ecrit:
> > Right, this would require a speech-dispatcher module file. The thing
> > I'm not sure about is how to do the serial port I/O from
> > speech-dispatcher. The speech-dispatcher modules are .conf files,
> 
> Not exactly. speech-dispatcher modules are .c programs, which can use a
> .conf file to be configured.
> 
> It happens that there is the generic module which can be made to run
> whatever command you want by configuring it in a .conf file.
> 
> But you can really do whatever you want in an entirely new .c speechd
> module.
> 
> Samuel
> _______________________________________________
> Speakup mailing list
> Speakup@linux-speakup.org
> http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup
> 

-- 
Your life is like a penny.  You're going to lose it.  The question is:
How do
you spend it?

         John Covici wb2una
         covici@ccs.covici.com

Re: driving a serial synthesizer through speech-dispatcher

[Samuel Thibault]

John Covici, le ven. 07 févr. 2020 04:41:12 -0500, a ecrit:
> You could use speechd-up and speakup would not have the device at all,
> so you could do what you like with it in your driver.

But then you are reimplementing a driver, while speakup already knows
how to drive them, it just needs a few ioctls to make the interface.

Samuel

Re: driving a serial synthesizer through speech-dispatcher

[John G Heim]

There would already be code in speakup to talk to different hardware 
synths. I don't know how portable that code is. But you could probably 
look at the speakup code to see how to set the speech rate for a 
liteTalk for example. If that code is portable, a large part of the work 
would already be done.
On 2/7/20 1:05 AM, John Covici wrote:
> You can treat the serial port like a file (this is linux after all),
> just do open in the i nit section and write characters to the file.
> The name will be something like /dev/ttyS0 or similar.  Note the
> capital S.  As an example look at the espeak  driver, so you get all
> the things in the right places.
> 
> 
> On Thu, 06 Feb 2020 23:29:43 -0500,
> Gregory Nowak wrote:
>>
>> Right, this would require a speech-dispatcher module file. The thing
>> I'm not sure about is how to do the serial port I/O from
>> speech-dispatcher. The speech-dispatcher modules are .conf files, so
>> maybe the answer would be a separate program to expose the serial port
>> to speech-dispatcher modules, or a modification to the actual
>> speech-dispatcher code to do that.
>>
>> What I like about the synth_direct approach is that speakup already
>> does the heavy lifting, and speakup and gnome-speech would use
>> speakup's interface to talk to the synth, which would mean they both
>> wouldn't be setting parameters and causing a mess. The disadvantage is
>> it requires speakup to be loaded, which I already use anyway, so it
>> doesn't bother me personally.
>>
>> Greg
>>
>>
>> On Thu, Feb 06, 2020 at 11:10:24PM -0500, John Covici wrote:
>>> You might need to write a driver in speech-dispatcher, that would do
>>> exactly what you want.  I have been thinking about this for years, but
>>> never had the time to do it.
>>
>>
>> -- 
>> web site: http://www.gregn.net
>> gpg public key: http://www.gregn.net/pubkey.asc
>> skype: gregn1
>> (authorization required, add me to your contacts list first)
>> If we haven't been in touch before, e-mail me before adding me to your contacts.
>>
>> --
>> Free domains: http://www.eu.org/ or mail dns-manager@EU.org
>> _______________________________________________
>> Speakup mailing list
>> Speakup@linux-speakup.org
>> http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup
>>
> 

Re: driving a serial synthesizer through speech-dispatcher

[Karen Lewellen]

My understanding from the first post  is that the goal is using serial 
synthesizers  in more graphical settings comparative to how orca functions.
Would not speekup present limitations for such a goal?



On Fri, 7 Feb 2020, John G Heim wrote:

> There would already be code in speakup to talk to different hardware synths. 
> I don't know how portable that code is. But you could probably look at the 
> speakup code to see how to set the speech rate for a liteTalk for example. If 
> that code is portable, a large part of the work would already be done.
> On 2/7/20 1:05 AM, John Covici wrote:
>>  You can treat the serial port like a file (this is linux after all),
>>  just do open in the i nit section and write characters to the file.
>>  The name will be something like /dev/ttyS0 or similar.  Note the
>>  capital S.  As an example look at the espeak  driver, so you get all
>>  the things in the right places.
>>
>>
>>  On Thu, 06 Feb 2020 23:29:43 -0500,
>>  Gregory Nowak wrote:
>> > 
>> >  Right, this would require a speech-dispatcher module file. The thing
>> >  I'm not sure about is how to do the serial port I/O from
>> >  speech-dispatcher. The speech-dispatcher modules are .conf files, so
>> >  maybe the answer would be a separate program to expose the serial port
>> >  to speech-dispatcher modules, or a modification to the actual
>> >  speech-dispatcher code to do that.
>> > 
>> >  What I like about the synth_direct approach is that speakup already
>> >  does the heavy lifting, and speakup and gnome-speech would use
>> >  speakup's interface to talk to the synth, which would mean they both
>> >  wouldn't be setting parameters and causing a mess. The disadvantage is
>> >  it requires speakup to be loaded, which I already use anyway, so it
>> >  doesn't bother me personally.
>> > 
>> >  Greg
>> > 
>> > 
>> >  On Thu, Feb 06, 2020 at 11:10:24PM -0500, John Covici wrote:
>> > >  You might need to write a driver in speech-dispatcher, that would do
>> > >  exactly what you want.  I have been thinking about this for years, but
>> > >  never had the time to do it.
>> > 
>> > 
>> >  -- 
>> >  web site: http://www.gregn.net
>> >  gpg public key: http://www.gregn.net/pubkey.asc
>> >  skype: gregn1
>> >  (authorization required, add me to your contacts list first)
>> >  If we haven't been in touch before, e-mail me before adding me to your 
>> >  contacts.
>> > 
>> >  --
>> >  Free domains: http://www.eu.org/ or mail dns-manager@EU.org
>> >  _______________________________________________
>> >  Speakup mailing list
>> >  Speakup@linux-speakup.org
>> >  http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup
>> >
>> 
> _______________________________________________
> Speakup mailing list
> Speakup@linux-speakup.org
> http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup
>
>

Re: driving a serial synthesizer through speech-dispatcher

[Gregory Nowak]

Thanks to everyone for all the input so far. I didn't realize that
some of the modules with the .conf files also have binaries as the
backend. I do indeed see that they're there in
/usr/lib/speech-dispatcher on my system. I'll have to look at the
speech-dispatcher code, and see how they do what they do.

I'm still in favor of letting speakup drive the serial synth. Thanks
to Samuel's comment below, I know what part of the speakup code to
look at, and the comment that it should be a fairly simple beginner kernel
hacking project is encouraging. I'll add this project to my to do list,
and will explore it once I have more time, hopefully sooner rather
than later. Thanks again.

Greg


On Fri, Feb 07, 2020 at 09:10:24AM +0100, Samuel Thibault wrote:
> Indeed. But there is another approach, which was made for this: using
> /dev/synth. For now that file only supports writing to it to get
> text emitted, but ioctls can be defined to drive it, such as setting
> pitch etc. in a portable way across synthesizers (speakup will handle
> these details), but also interrupting. That all happens in speakup's
> devsynth.c and can be a relatively easy task for kernel hacking
> beginners.
> 
> Samuel
> _______________________________________________
> Speakup mailing list
> Speakup@linux-speakup.org
> http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup

-- 
web site: http://www.gregn.net
gpg public key: http://www.gregn.net/pubkey.asc
skype: gregn1
(authorization required, add me to your contacts list first)
If we haven't been in touch before, e-mail me before adding me to your contacts.

--
Free domains: http://www.eu.org/ or mail dns-manager@EU.org

Re: driving a serial synthesizer through speech-dispatcher

[Gregory Nowak]

I forgot to add one more thing. The idea of using speechd-up to let
both speakup and gnome-speech use speech-dispatcher is not a bad
one. However, speechd-up was written as a proof of concept, and
doesn't implement full functionality like espeakup does. I suppose I
could look at that code too, and polish it to the level of espeakup,
but I frankly don't feel like it. A different module to drive a
different synth would still be required if speakup didn't do the heavy
lifting, instead of writing just one module.

Greg


On Fri, Feb 07, 2020 at 09:10:24AM +0100, Samuel Thibault wrote:
> > An even better approach is to use
> > /sys/accessibility/speakup/synth_direct for the speech output. This
> > would use speakup settings in use for the synthesizer. The problem
> > with this is the same as with the cat command; no way to interrupt
> > speech,
> 
> Indeed. But there is another approach, which was made for this: using
> /dev/synth. For now that file only supports writing to it to get
> text emitted, but ioctls can be defined to drive it, such as setting
> pitch etc. in a portable way across synthesizers (speakup will handle
> these details), but also interrupting. That all happens in speakup's
> devsynth.c and can be a relatively easy task for kernel hacking
> beginners.
> 
> Samuel
> _______________________________________________
> Speakup mailing list
> Speakup@linux-speakup.org
> http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup

-- 
web site: http://www.gregn.net
gpg public key: http://www.gregn.net/pubkey.asc
skype: gregn1
(authorization required, add me to your contacts list first)
If we haven't been in touch before, e-mail me before adding me to your contacts.

--
Free domains: http://www.eu.org/ or mail dns-manager@EU.org

Re: driving a serial synthesizer through speech-dispatcher

[John Covici]

The reason I got interested in speechd-up is that the espeak module
does not let you use the equivalent of indexing, as far as I know, to
test in a document try speakup-r and see if it works.

On Fri, 07 Feb 2020 18:02:58 -0500,
Gregory Nowak wrote:
> 
> I forgot to add one more thing. The idea of using speechd-up to let
> both speakup and gnome-speech use speech-dispatcher is not a bad
> one. However, speechd-up was written as a proof of concept, and
> doesn't implement full functionality like espeakup does. I suppose I
> could look at that code too, and polish it to the level of espeakup,
> but I frankly don't feel like it. A different module to drive a
> different synth would still be required if speakup didn't do the heavy
> lifting, instead of writing just one module.
> 
> Greg
> 
> 
> On Fri, Feb 07, 2020 at 09:10:24AM +0100, Samuel Thibault wrote:
> > > An even better approach is to use
> > > /sys/accessibility/speakup/synth_direct for the speech output. This
> > > would use speakup settings in use for the synthesizer. The problem
> > > with this is the same as with the cat command; no way to interrupt
> > > speech,
> > 
> > Indeed. But there is another approach, which was made for this: using
> > /dev/synth. For now that file only supports writing to it to get
> > text emitted, but ioctls can be defined to drive it, such as setting
> > pitch etc. in a portable way across synthesizers (speakup will handle
> > these details), but also interrupting. That all happens in speakup's
> > devsynth.c and can be a relatively easy task for kernel hacking
> > beginners.
> > 
> > Samuel
> > _______________________________________________
> > Speakup mailing list
> > Speakup@linux-speakup.org
> > http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup
> 
> -- 
> web site: http://www.gregn.net
> gpg public key: http://www.gregn.net/pubkey.asc
> skype: gregn1
> (authorization required, add me to your contacts list first)
> If we haven't been in touch before, e-mail me before adding me to your contacts.
> 
> --
> Free domains: http://www.eu.org/ or mail dns-manager@EU.org
> _______________________________________________
> Speakup mailing list
> Speakup@linux-speakup.org
> http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup
> 

-- 
Your life is like a penny.  You're going to lose it.  The question is:
How do
you spend it?

         John Covici wb2una
         covici@ccs.covici.com

Re: driving a serial synthesizer through speech-dispatcher

[John Covici]

Sorry instead of espeak I meant espeakup.

On Fri, 07 Feb 2020 18:02:58 -0500,
Gregory Nowak wrote:
> 
> I forgot to add one more thing. The idea of using speechd-up to let
> both speakup and gnome-speech use speech-dispatcher is not a bad
> one. However, speechd-up was written as a proof of concept, and
> doesn't implement full functionality like espeakup does. I suppose I
> could look at that code too, and polish it to the level of espeakup,
> but I frankly don't feel like it. A different module to drive a
> different synth would still be required if speakup didn't do the heavy
> lifting, instead of writing just one module.
> 
> Greg
> 
> 
> On Fri, Feb 07, 2020 at 09:10:24AM +0100, Samuel Thibault wrote:
> > > An even better approach is to use
> > > /sys/accessibility/speakup/synth_direct for the speech output. This
> > > would use speakup settings in use for the synthesizer. The problem
> > > with this is the same as with the cat command; no way to interrupt
> > > speech,
> > 
> > Indeed. But there is another approach, which was made for this: using
> > /dev/synth. For now that file only supports writing to it to get
> > text emitted, but ioctls can be defined to drive it, such as setting
> > pitch etc. in a portable way across synthesizers (speakup will handle
> > these details), but also interrupting. That all happens in speakup's
> > devsynth.c and can be a relatively easy task for kernel hacking
> > beginners.
> > 
> > Samuel
> > _______________________________________________
> > Speakup mailing list
> > Speakup@linux-speakup.org
> > http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup
> 
> -- 
> web site: http://www.gregn.net
> gpg public key: http://www.gregn.net/pubkey.asc
> skype: gregn1
> (authorization required, add me to your contacts list first)
> If we haven't been in touch before, e-mail me before adding me to your contacts.
> 
> --
> Free domains: http://www.eu.org/ or mail dns-manager@EU.org
> _______________________________________________
> Speakup mailing list
> Speakup@linux-speakup.org
> http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup
> 

-- 
Your life is like a penny.  You're going to lose it.  The question is:
How do
you spend it?

         John Covici wb2una
         covici@ccs.covici.com

Re: driving a serial synthesizer through speech-dispatcher

[Samuel Thibault]

John Covici, le ven. 07 févr. 2020 21:14:19 -0500, a ecrit:
> The reason I got interested in speechd-up is that the espeak module
> does not let you use the equivalent of indexing, as far as I know,

? It does have index marking support.

Samuel