(Note: I originally reported this as Firefox bug 1723319, but since it doesn't occur when I run Firefox without apulse, I now suspect it might be an apulse problem.)
Firefox has a text-to-speech feature that activates when a tab enters Reader View mode, even if nothing is being spoken. If Reader View is activated, any speaking applications launched from then on are silent. It seems that speech-dispatcher processes launched on behalf of Firefox enter a bad state, ignoring speech requests from other apps.
This condition persists until several seconds after Firefox and all other speech apps exit, when the speech-dispatcher process also exits.
The problem is avoided if some other speech app is the first to launch. For example, if I launch mumble before Firefox enters Reader View mode, both can then take turns speaking. The problem only appears if Firefox was the first to activate a speech-related feature.
I have tried telling apulse to use alternative ALSA devices (such as dmix and plughw devices) via the APULSE_PLAYBACK_DEVICE environment variable, but haven't found one that works for apulse without triggering the problem.
apulse 0.1.13 and 0.1.12
Firefox 90.0
Xubuntu 20.04.2 LTS
Tested with firefox, apulse+firefox, mumble, spd-say.
(Note: I originally reported this as Firefox bug 1723319, but since it doesn't occur when I run Firefox without apulse, I now suspect it might be an apulse problem.)
Firefox has a text-to-speech feature that activates when a tab enters Reader View mode, even if nothing is being spoken. If Reader View is activated, any speaking applications launched from then on are silent. It seems that speech-dispatcher processes launched on behalf of Firefox enter a bad state, ignoring speech requests from other apps.
This condition persists until several seconds after Firefox and all other speech apps exit, when the speech-dispatcher process also exits.
The problem is avoided if some other speech app is the first to launch. For example, if I launch mumble before Firefox enters Reader View mode, both can then take turns speaking. The problem only appears if Firefox was the first to activate a speech-related feature.
I have tried telling apulse to use alternative ALSA devices (such as dmix and plughw devices) via the APULSE_PLAYBACK_DEVICE environment variable, but haven't found one that works for apulse without triggering the problem.
apulse 0.1.13 and 0.1.12 Firefox 90.0 Xubuntu 20.04.2 LTS Tested with firefox, apulse+firefox, mumble, spd-say.