The start/end_listening sounds should be called with the pulse media.role=phone to enable ducking

[Ramblurr]

Copied from MycroftAI/mycroft-core#3148

This will likely never get fixed upstream since MycrodtAI has moved on to dinkum. But now that I am building my home setup around ovos, maybe there is a chance?

The linked ticket describes the issue.

Basically, the start listening sound play command should be passed the same environment as the tts play command.

This let's pulse properly handle ducking of the start listening sound.

Without this, as it currently stands, the start listening sound is played without the phone role, so it is treated as music. This is problematic because I want the start listening sound and the tts to be played at the same volume.

[JarbasAl]

the fix for this should be simple and go here https://github.com/OpenVoiceOS/ovos-utils/blob/dev/ovos_utils/sound/__init__.py

[emphasize]

Have to check os.environ. The way this is coded is weird and indicates the common PULSE_PROP environmental variable is media.role=phone

_ENVIRONMENT = deepcopy(os.environ)
_ENVIRONMENT['PULSE_PROP'] = 'media.role=music'


def _get_pulse_environment(config):
    """Return environment for pulse audio depeding on ducking config."""
    tts_config = config.get('tts', {})
    if tts_config and tts_config.get('pulse_duck'):
        return _ENVIRONMENT
    else:
        return os.environ

But the more likely interpretation is that the ducking media role is misunderstood (music ducks phone).