Software Speech Synth with Speakup -- Help, please.

Software Speech Synth with Speakup -- Help, please.

[Hugh Esco]

Greetings:

I put these questions a few days ago and the first one in a number of ways 
over a week ago but still have seen no answers that instill in me the 
confidence to move forward.  Any and all help is appreciated.

(1)     As I am so close on having Speakup / Festival / Middleware working 
on this machine I'm working on, any ideas what it will take to finish that 
install successfully?  (My progress to date is documented in the thread: 
"Any experience w/ Debian Woody Speakup Packages?")

(2)     If I were to install yasr, would it work with Speakup?  Would it 
announce kernel panics during boot-up?

(3)     I already have eflite 0.3.5-2 working with emacspeak on this box, 
installed from the .deb packages.  Do I still need to re-install it from 
source?  Might this break my emacspeak installation?  Can you tell me more 
about compiling eflite with the appropriate voices?  Where would I find 
those voices?

-- Hugh

At 08:09 AM 7/21/03 -0400, you wrote:
>Hi, Hugh.
>To get started using yasr with software speech all you need do is go to:
>http://yasr.sourceforge.net
>Download the yasr 0.6.4 source, and then grab flite and eflite source which
>there are links to.
>   You would build flite, and then compile eflite using the 16k voices not
>the 8k voices, and then set the synth port in yasr.conf, and you have
>speech.

Re: Software Speech Synth with Speakup -- Help, please.

[Owen Patrick Smith]

Unfortunately I haven't been following this list very closely lately and I
can't find the thread in the archives so if you could give me a summarry
of your status with the middleware program I'll try to help.

Owen Smith
ender3rd@wpi.edu

On Sun, 27 Jul 2003, Hugh Esco wrote:

> Greetings:
>
> I put these questions a few days ago and the first one in a number of ways
> over a week ago but still have seen no answers that instill in me the
> confidence to move forward.  Any and all help is appreciated.
>
> (1)     As I am so close on having Speakup / Festival / Middleware working
> on this machine I'm working on, any ideas what it will take to finish that
> install successfully?  (My progress to date is documented in the thread:
> "Any experience w/ Debian Woody Speakup Packages?")
>
> (2)     If I were to install yasr, would it work with Speakup?  Would it
> announce kernel panics during boot-up?
>
> (3)     I already have eflite 0.3.5-2 working with emacspeak on this box,
> installed from the .deb packages.  Do I still need to re-install it from
> source?  Might this break my emacspeak installation?  Can you tell me more
> about compiling eflite with the appropriate voices?  Where would I find
> those voices?
>
> -- Hugh
>
> At 08:09 AM 7/21/03 -0400, you wrote:
> >Hi, Hugh.
> >To get started using yasr with software speech all you need do is go to:
> >http://yasr.sourceforge.net
> >Download the yasr 0.6.4 source, and then grab flite and eflite source which
> >there are links to.
> >   You would build flite, and then compile eflite using the 16k voices not
> >the 8k voices, and then set the synth port in yasr.conf, and you have
> >speech.
>
>
>
> _______________________________________________
> Speakup mailing list
> Speakup@braille.uwo.ca
> http://speech.braille.uwo.ca/mailman/listinfo/speakup
>
>
>

Owen: Re: Software Speech Synth with Speakup -- Help, please.

[Hugh Esco]

Mr. Smith:

Thank you so much for your kind offer.  I've copied the relevant posts back 
to the list (in case others might help) under this new subject line.  My 
apologies for gumming of the list with these re-posts.

The one outstanding issue, I think is getting middleware to execute and 
scripting it in to the start-up routines.  The posts I sent though go back 
through the process and show how it is I got to where I am now.

-- Hugh Esco

At 12:39 PM 7/27/03 -0400, you wrote:
>Unfortunately I haven't been following this list very closely lately and I
>can't find the thread in the archives so if you could give me a summarry
>of your status with the middleware program I'll try to help.
>
>Owen Smith
>ender3rd@wpi.edu
>
>On Sun, 27 Jul 2003, Hugh Esco wrote:
>
> > Greetings:
> >
> > I put these questions a few days ago and the first one in a number of ways
> > over a week ago but still have seen no answers that instill in me the
> > confidence to move forward.  Any and all help is appreciated.
> >
> > (1)     As I am so close on having Speakup / Festival / Middleware working
> > on this machine I'm working on, any ideas what it will take to finish that
> > install successfully?  (My progress to date is documented in the thread:
> > "Any experience w/ Debian Woody Speakup Packages?")
> >
> > (2)     If I were to install yasr, would it work with Speakup?  Would it
> > announce kernel panics during boot-up?
> >
> > (3)     I already have eflite 0.3.5-2 working with emacspeak on this box,
> > installed from the .deb packages.  Do I still need to re-install it from
> > source?  Might this break my emacspeak installation?  Can you tell me more
> > about compiling eflite with the appropriate voices?  Where would I find
> > those voices?
> >
> > -- Hugh
> >
> > At 08:09 AM 7/21/03 -0400, you wrote:
> > >Hi, Hugh.
> > >To get started using yasr with software speech all you need do is go to:
> > >http://yasr.sourceforge.net
> > >Download the yasr 0.6.4 source, and then grab flite and eflite source 
> which
> > >there are links to.
> > >   You would build flite, and then compile eflite using the 16k voices not
> > >the 8k voices, and then set the synth port in yasr.conf, and you have
> > >speech.
> >
> >
> >
> > _______________________________________________
> > Speakup mailing list
> > Speakup@braille.uwo.ca
> > http://speech.braille.uwo.ca/mailman/listinfo/speakup
> >
> >
> >
>
>
>_______________________________________________
>Speakup mailing list
>Speakup@braille.uwo.ca
>http://speech.braille.uwo.ca/mailman/listinfo/speakup

Re: Software Speech Synth with Speakup -- Help, please.

[Thomas D. Ward]

Hi. sorry about the delay. I have been extremely busy, and haven't been on
the net much of this week.
Anyway, the answers to your questions are below. I have placed them in the
body of your message below each of your questions.

----- Original Message -----
From: Hugh Esco <hesco@greens.org>
To: <speakup@braille.uwo.ca>
Sent: Sunday, July 27, 2003 11:52 AM
Subject: Software Speech Synth with Speakup -- Help, please.


> Greetings:
>
> I put these questions a few days ago and the first one in a number of ways
> over a week ago but still have seen no answers that instill in me the
> confidence to move forward.  Any and all help is appreciated.

 >
> (1)     As I am so close on having Speakup / Festival / Middleware working
> on this machine I'm working on, any ideas what it will take to finish that
> install successfully?  (My progress to date is documented in the thread:
> "Any experience w/ Debian Woody Speakup Packages?")

I'm not sure how to answer this question as I tried the speakup/festival
thing when it first came out, but there were some things I didn't like about
it so I switched to yasr as a screen reader and  software speech solution.

>
> (2)     If I were to install yasr, would it work with Speakup?  Would it
> announce kernel panics during boot-up?

Well yasr is a completely different screen reader. It uses the flite,
festival lite, synth for software speech. Yasr   works by creating a sudo
shell over top of your existing shell.
Since yasr is not kernel based you will not be able to get kernel panics
etc. As I recall speakup can only get information such as kernel panics if
you are using a hardware synth like dectalk, doubletalk, etc... When you use
festival with speakup you can toss allot of the startup info right out of
the window.
That is the inharent problem with software speech. If you want all the power
of speakup such as kernel panics then you require  a hardware synth.
If you want speech without all the bells and wistles then yasr with flite or
speakup will get you along in a pinch but you loose the ability to install
Linux from scratch, reading kernel panics, all the startup screens, etc...

>
> (3)     I already have eflite 0.3.5-2 working with emacspeak on this box,
> installed from the .deb packages.  Do I still need to re-install it from
> source?  Might this break my emacspeak installation?  Can you tell me more
> about compiling eflite with the appropriate voices?  Where would I find
> those voices?

Well, I would suggest first by upgrading to eflite 0.37, and flite 1.2. I do
have yasr and eflite rpms now, but no debian packages so I suppose unless
alien can convert them you'd have to recompile from scratch.
I typically install yasr, flite, and eflite in /usr, and that is where all
my libs are linked in my rpms in /usr/lib.
So if you attempt to use alien on these rpms make sure the packages can
locate your libs.
Hth.

Re: Owen: Re: Software Speech Synth with Speakup -- Help, please.

[Christopher Moore]

Hello Hugh and others,
Well, I stalled everything according to the documentation, rebooted,
started the festival server and middleware, and actually got speech.  

The major concern I have is that festival is way behind anything I am try
to do.  This may be partially due to the slowness of festival.  The second
problem I'm having is that festival doesn't speak when I use the keypad.
This may be something I'm not doing although I checked the option to use
speakup as the default keymap.  

The next step may be to look at using flite instead of festival because it
should improve the response time.  

This is a great start and the gang at WPI is to be congratulated for moving
speakup into the software synth age.  I'm wondering whether middleware
could be rewritten to use eflite or some of the other emacspeak servers
such as viavoice and software dectalk.

Chris

Re: Owen: Re: Software Speech Synth with Speakup -- Help, please.

[Hugh Esco]

Please tell me what you did to get middleware and Festival working.  Are 
you working in Debian, by any chance?  I could not get the middleware 
executable to execute.  I'm clueless as to how it ought to interact with 
Festival and the Speakup kernel.  ANd would also appreciate some help on 
how to build it into the start-up scripts in /etc/init.d/.

-- Hugh

At 05:27 PM 7/28/03 -0400, you wrote:
>Hello Hugh and others,
>Well, I stalled everything according to the documentation, rebooted,
>started the festival server and middleware, and actually got speech.
>
>The major concern I have is that festival is way behind anything I am try
>to do.  This may be partially due to the slowness of festival.  The second
>problem I'm having is that festival doesn't speak when I use the keypad.
>This may be something I'm not doing although I checked the option to use
>speakup as the default keymap.
>
>The next step may be to look at using flite instead of festival because it
>should improve the response time.
>
>This is a great start and the gang at WPI is to be congratulated for moving
>speakup into the software synth age.  I'm wondering whether middleware
>could be rewritten to use eflite or some of the other emacspeak servers
>such as viavoice and software dectalk.
>
>Chris
>
>_______________________________________________
>Speakup mailing list
>Speakup@braille.uwo.ca
>http://speech.braille.uwo.ca/mailman/listinfo/speakup

Re: Owen: Re: Software Speech Synth with Speakup -- Help, please.

[Christopher Moore]

Hugh,
I'm not using debian, but the same techniques should apply.  Before you
worry about the startup script, you need to get it working by hand.

1.  Bot with your speakup kernel and festival driver.  I have a separate
section in my lilo.conf so I can type in "festival" at the boot prompt to
force speakup toload the usrdev driver.

2.  Once you log in as root, you need to make sure your soundcard is
functioning.  Try playing a wave file or and mp3 to make sure you have
sound enabled.

3.  Make sure that the festival modules are in your path.  My festival was
built in /opt directory so I used the following command to fix the PATH
variable.

export PATH=/opt/festival/bin:$PATH

4.  start the festival server as a background process.
% festival_server &

5. If you don't get any error messages, your're probably ready to run
middleware.  

% cd /path/to/middleware
% ./middleware &

If all goes well you should hear some speech.  Give us a report.

Chris

Re: Owen: Re: Software Speech Synth with Speakup -- Help, please.

[Hugh Esco]

Chris Moore:

Thanks for the blow by blow.  Here is what happened.

At 09:43 AM 7/29/03 -0400, you wrote:
>Hugh,
>I'm not using debian, but the same techniques should apply.  Before you
>worry about the startup script, you need to get it working by hand.
>
>1.  Bot with your speakup kernel and festival driver.  I have a separate
>section in my lilo.conf so I can type in "festival" at the boot prompt to
>force speakup toload the usrdev driver.

My default kernel in the lilo boot menu is Linux_speakup4.  That's handled.

>2.  Once you log in as root, you need to make sure your soundcard is
>functioning.  Try playing a wave file or and mp3 to make sure you have
>sound enabled.

I was able to use esdplay to play a sound
from /usr/share/sounds, both as root and as hesco.

>3.  Make sure that the festival modules are in your path.  My festival was
>built in /opt directory so I used the following command to fix the PATH
>variable.
>
>export PATH=/opt/festival/bin:$PATH

I have done this both as root and as hesco.
Only my Festival path is, I believe, /usr/lib/festival.

>4.  start the festival server as a background process.
>% festival_server &

This generates "command not found" bash errors.
In /usr/bin, I have festival and festival_client.
I can sort of successfully run festival &
It gives me a process id, then at the next prompt,
a start-up message and copyright notice,
but no festival prompt.  When I hit return,
it stops the background process and returns me
to a bash prompt.

I already have festival invoked by /etc/init.d/, however,
so this may not be an issue.  In fact, ps aux | grep festival
says I have three processes running:
         244     /usr/bin/festival --server
         1895    festival
         1898    grep festival

>5. If you don't get any error messages, your're probably ready to run
>middleware.
>
>% cd /path/to/middleware
>% ./middleware &

That would be:
         cd /usr/src/speakup-1.5/middleware
         ./middleware &

I get a process id (1902) and a message printed
at the next bash prompt that it:
         Failed to open file:  No such device

Hitting return gives me an Exit message.
ps aux | grep middleware shows only one line:
         1905    grep middleware

>If all goes well you should hear some speech.  Give us a report.
>
>Chris

And no speech.  What device is it that middleware is unable to open?
Might this be a permission problem?  I've run this as root.
Should I try it as hesco?  That gives me the same result.

And this is where I was stuck last time?
I'm very close, but not yet there.
All clues are appreciated.  Thanks.

-- Hugh Esco

Re: Owen: Re: Software Speech Synth with Speakup -- Help, please.

[Christopher Moore]

Hugh,

I was surprised you didn't have a festival_server in your festival/bin
directory.  My version of festival is 1.4.3 which I built from sources last
week.  Perhaps running festival with the --server option is the same as
running festival_server, but I'd need to read more docs to verify it.  

You need festival to run as a server because that's how middleware
communicates with it.  I just look at the festival help screen and is says
you can invoke it as you did to go into server mode.  

Did you make the /dev/usrdev devicefile with the mknod command?  Do a 
% ls -l /dev/usrdev
Mine looks like:
crw-r--r--    1 root     root     252,   0 Jul 28 16:18 /dev/usrdev

You can check if you've loaded the festival speakup driver by doing an 
% ls /proc/speakup

You should see a subdirectory for the festival driver.  I don't recall its
name, but it should be the only subdirectory listed.  Within it are
settings for the festival driver.  

If I think of anything else, I'll pass it along.

Chris

Re: Owen: Re: Software Speech Synth with Speakup -- Help, please.

[Christopher Moore]

Hugh,
One more things.  I noticed your message from middleware:
Failed to open file:  No such device

This is the exact message I get when I don't have the usrdev driver running
with speakup.  
Be sure your lilo.conf entry for your boot image has an append line like:
apend="speakup_synth=usrdev"

Chris

Re: Owen: Re: Software Speech Synth with Speakup -- Help, please.

[Hugh Esco]

Chris:

Here is what I got in response to those first two questions:

hesco@biko:~$ ls -l /dev/usrdev
crw-r--r--    1 root     root     252,   0 Jul 19 22:54 /dev/usrdev
hesco@biko:~$ ls -l /proc/speakup
total 0
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 attribute_bleep
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 bell_pos
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 bleep_time
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 characters
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 key_echo
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 line_wrap_bleep
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 no_interrupt
-r--r--r--    1 root     root            0 Jul 30 20:39 punc_all
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 punc_level
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 punc_most
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 punc_some
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 say_control
--w--w--w-    1 root     root            0 Jul 30 20:39 silent
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 spell_delay
-rw-rw-rw-    1 root     root            0 Jul 30 20:39 synth
-r--r--r--    1 root     root            0 Jul 30 20:39 version
hesco@biko:~$

So, no, there is no directory in there.

Why not?  What would build that directory there?

I just edited lilo.conf to add a line:
         append="speakup_synth=usrdev"
to my Speakup stanza.

I assume I should spell it "append", not "apend" as you did?   I ran lilo 
without errors.  I'm rebooting now.  We'll soon see.

Well, it booted up with no speech, but then again, the last set of changes 
I made were handled manually and were not scripted in, so I'll repeat those 
and see what I can get.

I went back and re-examined my notes from the last round.  I re-tested 
sound, re-exported the path, used ps to check and make sure festival 
--server was still running, then attempted again to invoke 
middleware.  Again, I get "Failed to open file:  No such device" error.

I get the same results when I try all that as root, too.

What next?  Anyone?

-- Hugh

At 02:25 PM 7/30/03 -0400, you wrote:
>Hugh,
>
>I was surprised you didn't have a festival_server in your festival/bin
>directory.  My version of festival is 1.4.3 which I built from sources last
>week.  Perhaps running festival with the --server option is the same as
>running festival_server, but I'd need to read more docs to verify it.
>
>You need festival to run as a server because that's how middleware
>communicates with it.  I just look at the festival help screen and is says
>you can invoke it as you did to go into server mode.
>
>Did you make the /dev/usrdev devicefile with the mknod command?  Do a
>% ls -l /dev/usrdev
>Mine looks like:
>crw-r--r--    1 root     root     252,   0 Jul 28 16:18 /dev/usrdev
>
>You can check if you've loaded the festival speakup driver by doing an
>% ls /proc/speakup
>
>You should see a subdirectory for the festival driver.  I don't recall its
>name, but it should be the only subdirectory listed.  Within it are
>settings for the festival driver.
>
>If I think of anything else, I'll pass it along.
>
>Chris
>_______________________________________________
>Speakup mailing list
>Speakup@braille.uwo.ca
>http://speech.braille.uwo.ca/mailman/listinfo/speakup

Re: Owen: Re: Software Speech Synth with Speakup -- Help, please.

[Christopher Moore]

Hugh,
I'm still not sure whether you're loading the usrdev driver.  
In your /usr/src/linux/.config file you should have the following line:
CONFIG_SPEAKUP_USRDEV=y

Also, the line directly below that defines the default speakup driver;
CONFIG_SPEAKUP_DEFAULT="none"  (in my case).

Why don't you send your lilo.conf.  I'd like to have a look at it.  

Chris