ASR Fallback seems not installed before it is required?

[philipp2310]

As it seems the fallback ASR is installed only when it is used - most of the time it should be used when the internet was dropped -> but how to install in that case?

[Psychokiller1888]

That's.... true.... :)

[philipp2310]

and a second thing: Alice changes to the fallback ASR when the internet connection is lost, even if the main ASR doesn't require internet

[Psychokiller1888]

	def onInternetLost(self):
		if self._asr.isOnlineASR:
			self.logInfo('Internet lost, switching to offline Asr')
			self.restartEngine()

I don't get why she would switch?

[philipp2310]

Ha, yeah, you are right, it is not on loss of internet but on reestablishing:

			self.logInfo('Connected to internet, switching Asr')
			self.restartEngine()

a restart is forced even though no onlineASR or change is required.
Easy to solve as user by setting the Fallback == the main STT -> there currently isn't a reason to use the fallback apart from a failing start of the main ASR (model missing, training not consistent).

I'll adjust it so you can have a real fallback when the main ASR crashes - without relevance of the internet (e.g. Coqui as "hot" software with potential changes, with deepspeech 6.1 as a solid fallback that won't ever crash)

Edit: Even easier: if both ASR are offline ASR, the "keepASR offline" should be set!