feat/listening_modes

[NeonJarbas]

adds ListenMode, this is globally set in mycroft.conf and can be one of the following

• WAKE_WORD - the classic mode, listen for wakewords and trigger actions based on them
• CONTINUOUS - use VAD to detect speech and always transcribe it, ignore STT failures
  • ⚠️ offline streaming STT highly recommended! ⚠️
• HYBRID - after a wake word detection switch to continuous mode for X seconds (45 by default)
  • this allows follow up question without wake word, but good AEC is needed or mycroft will hear its own TTS

adds ListeningState, this is a runtime state and can be one of the following

• WAKE_WORD - currently waiting for a wake word
• CONTINUOUS - currently using VAD to detect speech
• RECORDING - save all audio until told to stop, then reset to default mode
  • think "captain's log, stardate 184876, today the warp core exploded, bla bla bla"

other notes:

• bus events can be used to query and change the listening mode / state
• new class of hotwords, "stopwords", only used while in RECORDING state to stop recording
• wakeword/utterance save paths now respect XDG
• added filename to message.data in "recognizer_loop:record_end"

[codecov]

Codecov Report

Merging #150 (1f5fb2c) into dev (6ceb058) will increase coverage by 3.24%.
The diff coverage is 36.58%.

@@            Coverage Diff             @@
##              dev     #150      +/-   ##
==========================================
+ Coverage   50.35%   53.59%   +3.24%     
==========================================
  Files         119      152      +33     
  Lines       10077     9725     -352     
==========================================
+ Hits         5074     5212     +138     
+ Misses       5003     4513     -490     

Impacted Files	Coverage Δ
mycroft/audio/__main__.py	0.00% <0.00%> (ø)
mycroft/client/enclosure/__main__.py	0.00% <0.00%> (ø)
mycroft/client/enclosure/mark1/arduino.py	0.00% <0.00%> (ø)
mycroft/client/enclosure/mark1/eyes.py	0.00% <0.00%> (ø)
mycroft/client/enclosure/mark1/mouth.py	0.00% <0.00%> (ø)
mycroft/client/speech/__main__.py	0.00% <0.00%> (ø)
mycroft/client/speech/hotword_factory.py	0.00% <0.00%> (-88.89%)	⬇️
mycroft/client/speech/service.py	0.00% <0.00%> (ø)
mycroft/client/speech/silence.py	0.00% <0.00%> (-42.86%)	⬇️
mycroft/client/text/__init__.py	0.00% <0.00%> (ø)
... and 138 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[NeonDaniel]

Any idea why codecov is annotating unmodified lines? Makes the PR hard to read...

[NeonDaniel]

https://github.com/OpenVoiceOS/ovos-core/pull/150/files#diff-0f7600a790272b32edf283b7755335cab5e9458d57eae200a4aa6648d43ed272L483

self._listen_triggered should be set to False here to avoid an infinite listening loop

[NeonDaniel]

Tested in NeonGeckoCom/neon_speech#96