[Tts-project] Bug#948679: Bug#948679: speech-dispatcher: doesn't start, first due to /var/run/…/cache/… then Can't bind local socket

[Samuel Thibault]

Thorsten Glaser, le dim. 12 janv. 2020 16:31:42 +0100, a ecrit:
> > There is no need to start speech-dispatcher as root. Nowadays you can
> 
> ah, okay. It did try to start it system-wide during package installation,

??
That's very surprising, we install it with --no-enable --no-start

And the installation log on a freshly installed box confirms that
default configuration:

insserv: warning: current start runlevel(s) (empty) of script `speech-dispatcher` overrides LSB defaults (2 3 4 5).
insserv: warning: current stop runlevel(s) (0 1 2 3 4 5 6) of script `speech-dispatcher` overrides LSB defaults (0 1 6).

> So, we have the startup issues addressed, I think. Now, no sound,
> although this might be a separate bugreport?

As mentioned previously it can also be simply that orca isn't actually
looking at the right place. Thus first testing with spd-say.

> > spd-say "foo"
> >
> > and if that doesn't work, check out the log files in
> > /run/user/1000/speech-dispatcher/log
> 
> That path does not exist, but strace -e file showed there are logs in
> /tmp/runtime-tglase/speech-dispatcher/log/

Ok, so that path varies with environments...

> which I attached (as gzipped
> tarball, although probably only speech-dispatcher.log has meaningful
> informatioin).
> 
> The logs contain info from exactly one spd-say invocation.

Ok, thanks.

> strace shows successful openat() of these sound-related devices:
> /dev/snd/controlC0 /dev/snd/pcmC0D0p /dev/snd/timer

Yes. The log shows two things:

- only the dummy, generic, and mary-generic modules get enabled. You
  need to let apt install the recommended speech-dispatcher-espeak-ng
  package, in order to get at least one speech output backend.

- the generic module gets to be recognized as "working" while it
  shouldn't even appear. speech-dispatcher thus doesn't default to dummy
  which would have told you of the situation. I'll have a look at that.

(also, make sure in your mixer that the alsa output is not muted).

Samuel