Not exactly this, but iOS and google both have voice mode shortcuts you can use "Hey Siri" (which i think only works when the device is plugged in) and "Ok, Google".

The problem I see is users getting frustrated with false positives - having the device do things they are not expecting. Pretty much the experience someone has when they first try vi or ed.

They still require the utterly ridiculous "Ok, Google" activation though. I know its irrational, but I really don't want to have to use phrases I would never use in day to day life to communicate with my computer.

Just wonder whenever someone had experimented with use of constructed or just uncommon languages for such purposes. Such approach is certainly not consumer-friendly, but fun nonetheless.

Say, a geeky enough person could use Esperanto or Lojban to control always-listening home automation without worrying too much it'll accidentally get unintended commands. Or, maybe, not a general purpose conlang, but a specially constructed one tailored just to quickly express the necessary concepts to machine.

Isn't that the whole point of an obscure catchword? To avoid the computer intercepting normal conversation?

Sure, but that's just an artifact of the crude capabilities of current speech recognition. A sufficiently advanced speech recognition engine wouldn't need you to insert a stopword to know that you're talking to it. It would just... know.

Sometimes I, as a human being, don't know if someone is talking to me unless they preface their statement with my name. How are computers supposed to be any better?

Long term, because it can monitor way more world state and be designed with as much processing power as necessary; you have one brain and don't want to spend 100% of your time working out if someone is talking to you.

That's a bit strange. I don't always know whether people are talking to me, and I wouldn't say my speech recognition is crude.

Why don't they just allow for naming a device? You'd use names in your everyday conversation.