[Emacspeak] TTS Problems with Pipewire and Fedora 37

[Tim Cross <theophilusx@xxxxxxxxxxx>]

Up until a few days ago, I've been running Emacspeak with no issues on
Fedora (initially with Fedora 36 and now Fedora 37). Unlike
Debian/Ubuntu based systems, Fedora moved to use pipewire and
wireplumber (session manager) for audio some time ago. Debian and Ubuntu
distributions are still in transition.

Last week or so, Fedora updated pipewire to 0.3.61 and since then, I've
had problems getting things to work. When I run the espeak server from
the command line, I hear the initial speech "espeak 1.5 1.1 espeak"
spoken as normal. However, if I then attempt to get the server to speak
more text by manually queueing up text and sending it, I get nothing
i.e.

q "This is a test"
d

This seems to be an issue with pipewire-pulse and the configuration used
with the espeak server. However, I'm very much out of my depth here. If
I restart pipewire and start Emacs immediately (it would seem before
pipewire-pulse is ready), I get a working server using pipewire-alsa.

Could the issue be something relating to all the configuration stuff
Raman has been doing with pulse? My guess is no as it is affecting the
server when I just run espeak on the command line from the servers
directory. I'm assuming at this level, it doesn't know anything about
sinks and should just be using the defaults.

Note that under pipewire, you cannot just drop a default.pa file into
the .config/pulse directory. You have to load the ladspa modules in the
pipewire-pulse.conf file in .config/pipewire and it seems to require
some additional configuration in order to get things working correctly
(something to do with filter chains I don't understand).

I also suspect, from what I'm learning, that once things have
transitioned to pipewire, there will be  a much more straight-forward way
to make the sort of sound tweaks Raman has been doing with pulse (the
ability to patch/route sound via different modules etc with pipewire
seems to be one of its main selling points).

To begin with, I just want to get a basic espeak TTS working with the
default sink, no remapping, no additional ladspa module - just a clean
pipewire base config to start with. If anyone else has made progress
here, would be very interested in what you have done as I'm very much
out of my depth here.