How to disable speech-dispatcher in Ubuntu 20.04?

Question

I've got some problems with the recognition of my audio devices (see here for more detail). One suggested fix that floats around the internet is to disable the speech-dispatcher, by changing the following line in the configuration file /etc/default/speech-dispatcher

RUN=yes

to

RUN=no

However, the file /etc/default/speech-dispatcher does not exist on my Ubuntu 20.04. I tried creating it with the proposed line, but I didn't recognize any different behaviour, I was able to use the speech dispatcher with spd-say.

How do I check if adding the configuration file with RUN=no has any effect? And if it does not work this way in Ubuntu 20.04, how can I disable the autostart of speech-dispatcher?

Any help would be appreciated :)

Answer 1

After many different fixes, some that messed up audio after suspend I found that it was Firefox calling it I disabled speech-dispatcher in "about:config" and no more problem. The #1 answer will work, but I think it's better to stop it at the source.

Edit with instructions

The instructions below are valid as of 2023-09-12 on Fierefox version 117.0

1. Open a new Firefox tab and go to about:config, and press on Accept risk and continue.
2. Search for media.webspeech.synth.enabled and disable it.
3. Log out and log back in. The issue should be resolved.

Answer 2

Have you tried to uncomment the DisableAutoSpawn line in the file /etc/speech-dispatcher/speechd.conf, thus the server will not start automatically on requests from the clients ?

It looks like the speech-dispatcher service is not a system service started systematically like other SysV/systemd service, but launched on demand with your UID.

Answer 3

Speech Dispatcher will be idle usually.

Disabling the Accessibility - Screen Reader would stop Firefox (or any other) calling the Speech Dispatcher.

Answer 4

Background

There are a few answers floating around on how to disable speech dispatcher. I tried a few without success. Since I was in the process of writing a music player and constantly seeing speech-dispatcher as a sound source during testing it was annoying.

Below is a bash solution using the console and a python solution using a GUI.

From the Console

Enter these commands in the console after the $ prompt. The # hashtag comments are not entered.

$ uptime
 17:37:18 up 18 min,  1 user,  load average: 1.63, 1.67, 1.41
# Had rebooted three times: soft, warm and cold.
# The unwanted jobs didn't appear until 18 minutes after
# cold boot.


$ ps aux | grep speech
rick      8919  0.0  0.0 289032  5000 ?        Sl   17:33   0:00 /usr/lib/speech-dispatcher-modules/sd_dummy /etc/speech-dispatcher/modules/dummy.conf
rick      8922  0.0  0.0 326656  7468 ?        Sl   17:33   0:00 /usr/lib/speech-dispatcher-modules/sd_espeak /etc/speech-dispatcher/modules/espeak.conf
rick      8927  0.0  0.0 289044  5028 ?        Sl   17:33   0:00 /usr/lib/speech-dispatcher-modules/sd_generic /etc/speech-dispatcher/modules/generic.conf
rick      8930  0.0  0.0 289032  5032 ?        Sl   17:33   0:00 /usr/lib/speech-dispatcher-modules/sd_cicero /etc/speech-dispatcher/modules/cicero.conf
rick      8934  0.0  0.0  98768  2380 ?        Ssl  17:33   0:00 /usr/bin/speech-dispatcher --spawn --communication-method unix_socket --socket-path /run/user/1000/speech-dispatcher/speechd.sock
rick     17823  0.0  0.0  15776   928 pts/22   S+   17:37   0:00 grep --color=auto speech
# Above are four unwanted sound sinks and controller


$ pgrep -f speech-dispatcher/modules
8919
8922
8927
8930
# pgrep shows just the unwanted sound sinks' PID


$ pkill -f speech-dispatcher/modules
# pkill nukes the unwanted sound sinks by PID


$ ps aux | grep speech
rick      8934  0.0  0.0  98768  2380 ?        Ssl  17:33   0:00 /usr/bin/speech-dispatcher --spawn --communication-method unix_socket --socket-path /run/user/1000/speech-dispatcher/speechd.sock
rick     26836  0.0  0.0  15776   996 pts/22   S+   17:41   0:00 grep --color=auto speech
# Running ps again shows success

You can put two commands above into a bash script:

pgrep -f speech-dispatcher/modules
pkill -f speech-dispatcher/modules

The two line bash script that will delete the sound sinks for you.

---

Over-killing the PID Killing

The current project is a music player in Python so why not have it check when the speech-dispatcher gremlins show up? Then it displays a message:

Here's the python code to serve as a guide:

def check_speech_dispatcher(self):
    """ Four annoying speech dispatchers appear in Ubuntu """
    if not DELETE_SPEECH:
        return  # Already done or don't want to kill pids
found_pids = list()
for Sink in pav.sinks_now:
    if Sink.name == "speech-dispatcher":
        found_pids.append(Sink.pid)

if len(found_pids) == 0:
    return

global DELETE_SPEECH
DELETE_SPEECH = False  # Don't show message again this session

title = "Speech Dispatcher Jobs Discovered."
text = str(len(found_pids)) + " instance(s) of Speech"
text += "Dispatcher have been found.\n\n"
text += "Do you want to cancel the job(s)?\n"  # centered: \t breaks
answer = message.AskQuestion(self.play_top, title, text, 'no',
                             thread=self.get_refresh_thread())
text += "\n\t\tAnswer was: " + answer.result
self.info.cast(title + "\n\n" + text)

if answer.result != 'yes':
    return  # Don't delete pids
for pid in found_pids:
    ext.kill_pid_running(pid)

Near the bottom is info.cast() function. It zooms down a broadcast message. It's kind of like print() command for a GUI application:

Answer 5

In my case, pulseaudio might be the culprit that calls speech-dispatcher, not Firefox. I'd followed the suggestion to set Firefox's media.webspeech.synth.enabled to "false" but the problem persisted. However, killing the pulseaudio process was followed by a disappearance of speech-dispatcher.