Janina Sajka
2018-07-05 17:29:33 UTC
I'm reposting an excerpt from a message I posted on the Orca list this
morning, but it's really not about Orca. It's all about Speakup using
espeakup-git and espeak-ng.
Here's the context. Didier helped me to get both espeak-ng and
espeakup-git to build. I'm now able to launch it on my X86_64 Arch
system. But, there are very significant issues that will probably have
me rolling back to the old espeak and the shipping espeakup.
Details below quoted and edited to include only Speakup related info.
Any suggestions most welcome.
applications like espeak-ng, speech-dispatcher-git, espeakup-git, or
orca-git provide an easy mechanism for this situation. One is prompted
to choose editing the build and install files early in the process, so I
was able to add your code immediately above the make } directive in the
build file and all worked as expected from that point.
However, espeakup-git is not all that usable. I will leave this
configuration working for now in case someone has some good suggestions.
Problems noted so far:
1. The post installation guidance printed to screen by espeakup-git
install is not working for me. Enabling ***@user does nothing. In
order to get speech output, I have to enable espeakup explicitly, and
sometimes run a 'sudo systemctl restart espeakup' after logging in.
2. I discovered systemctl start speakupconf is assuming an
/etc/speakup directory with relevant data. I had to puruse journalctl to
discover this. Shouldn't this be checked on speakupconf install?
3. espeakup-git is honoring the default card directive in
* /etc/asound.conf. This is new. The shipping espeakup, as all
* previous incarnations, simply chose the first available audio
* card. I need my Speakup device to be different from the alsa
* default. This is a problem for me, though I can live with it for
* the moment.
4. Setting /etc/asound.conf to default card 0 loads a totaly
* unusable speakup. Screen review on the numpad is quite broken,
* especially 4-5-6 and 1-2-3 (word and char) review. Also, speech
* is NOT immediately silenced by Keypad-Enter or Ctrl. CARD 0 is
* the onboard HDA Intel audio device.
5. Resetting the default in /etc/asound.conf to CARD 2 fixes much
* of the problem in 4. above. CARD 2 is a Sennheiser headset
* device that I intend for making SIP calls, but both linphone and
* freeswitch are currently broken on Arch, so this is the
* configuration I've temporarily resumed using. Screen review
* works as expected here, but silencing speech is still an issue.
* Also, a spurious "space" is read with 7-8-9 and after pressing
* Enter on a command.
6. In both cases, both speech-dispatcher and espeakup, my cards are
blocked to other output, i.e. no dmix. This is new. The shipping
espeakup was not giving me this problem.
7. I should note that I have espeak-ng on two machines. The
problems resolved in this thread were from my main desktop X86_64
system. I have previously successfully installed Arch toa USB3 flash
drive for booting a 2013 vintage Apple Airbook. Until the update just
this week to the latest Arch 4.17.3 kernel, plus whatever audio drivers
came with that update, the problems with espeakup-git mentioned above
were NOT visible on the Apple, except for the reading of the spurious
"space" char. However, the Apple is similarly broken now. Something in
very recent updates has broken things badly on the Apple.
Janina
morning, but it's really not about Orca. It's all about Speakup using
espeakup-git and espeak-ng.
Here's the context. Didier helped me to get both espeak-ng and
espeakup-git to build. I'm now able to launch it on my X86_64 Arch
system. But, there are very significant issues that will probably have
me rolling back to the old espeak and the shipping espeakup.
Details below quoted and edited to include only Speakup related info.
Any suggestions most welcome.
...
Reinstall espeak-ng, then either edit the make command
https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=espeakup-git
si it becomes
make LDLIBS='-lespeak-ng -lpthread'
or edit the Makefile in the SLKBUILD,
inserting before the "make" line, in the build() function these lines
sed "/LDLIBS/s/.*/LDLIBS=-lespeak-ng -lpthread/" Makefile > dummy
mv dummy Makefile
I don't know what is the recommended Arch way, sorry I never
used Arch.
The Arch tools that allow for building Arch User Repository (AUR)Speakup won't work, and espeakup-git won't build with a similar error,
i.e. -lespeak not found.
Not a big issue.i.e. -lespeak not found.
Reinstall espeak-ng, then either edit the make command
https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=espeakup-git
si it becomes
make LDLIBS='-lespeak-ng -lpthread'
or edit the Makefile in the SLKBUILD,
inserting before the "make" line, in the build() function these lines
sed "/LDLIBS/s/.*/LDLIBS=-lespeak-ng -lpthread/" Makefile > dummy
mv dummy Makefile
I don't know what is the recommended Arch way, sorry I never
used Arch.
applications like espeak-ng, speech-dispatcher-git, espeakup-git, or
orca-git provide an easy mechanism for this situation. One is prompted
to choose editing the build and install files early in the process, so I
was able to add your code immediately above the make } directive in the
build file and all worked as expected from that point.
However, espeakup-git is not all that usable. I will leave this
configuration working for now in case someone has some good suggestions.
Problems noted so far:
1. The post installation guidance printed to screen by espeakup-git
install is not working for me. Enabling ***@user does nothing. In
order to get speech output, I have to enable espeakup explicitly, and
sometimes run a 'sudo systemctl restart espeakup' after logging in.
2. I discovered systemctl start speakupconf is assuming an
/etc/speakup directory with relevant data. I had to puruse journalctl to
discover this. Shouldn't this be checked on speakupconf install?
3. espeakup-git is honoring the default card directive in
* /etc/asound.conf. This is new. The shipping espeakup, as all
* previous incarnations, simply chose the first available audio
* card. I need my Speakup device to be different from the alsa
* default. This is a problem for me, though I can live with it for
* the moment.
4. Setting /etc/asound.conf to default card 0 loads a totaly
* unusable speakup. Screen review on the numpad is quite broken,
* especially 4-5-6 and 1-2-3 (word and char) review. Also, speech
* is NOT immediately silenced by Keypad-Enter or Ctrl. CARD 0 is
* the onboard HDA Intel audio device.
5. Resetting the default in /etc/asound.conf to CARD 2 fixes much
* of the problem in 4. above. CARD 2 is a Sennheiser headset
* device that I intend for making SIP calls, but both linphone and
* freeswitch are currently broken on Arch, so this is the
* configuration I've temporarily resumed using. Screen review
* works as expected here, but silencing speech is still an issue.
* Also, a spurious "space" is read with 7-8-9 and after pressing
* Enter on a command.
6. In both cases, both speech-dispatcher and espeakup, my cards are
blocked to other output, i.e. no dmix. This is new. The shipping
espeakup was not giving me this problem.
7. I should note that I have espeak-ng on two machines. The
problems resolved in this thread were from my main desktop X86_64
system. I have previously successfully installed Arch toa USB3 flash
drive for booting a 2013 vintage Apple Airbook. Until the update just
this week to the latest Arch 4.17.3 kernel, plus whatever audio drivers
came with that update, the problems with espeakup-git mentioned above
were NOT visible on the Apple, except for the reading of the spurious
"space" char. However, the Apple is similarly broken now. Something in
very recent updates has broken things badly on the Apple.
Janina