I have a Voice PE in my living room but it’s not right next to my couch. The Voice is configured to use the “Hey Jarvis” wake work. When I’m watching TV in that room the Voice will be accidentally triggered by the dialog on the TV at least once an hour and I don’t know how to tell the Voice to stop listening to the current request once it gets triggered. I can’t even press the button on the Voice because it’s on the other side of the room and I’m too lazy to run over to it.
I do have an automation set up which listens for sentences like “Stop”, “Cancel”, and “Never mind” and will then call the “Conversation: Process” action without any text to cancel the input. That automation works well if it hears those sentences right after being activated (like if I say “Hey Jarvis…cancel”) but doesn’t get triggered if it hears any dialog from the TV before I start yelling those commands at it.
When this happens I end up having to pause the show that I’m watching so that it doesn’t pick up any additional dialog and then wait for OpenAI to come back with some answer to a question based on whatever dialog it already heard.
For example, the Voice will somehow hear “Hey Jarvis” which I’m watching Monty Python and then we’re off to the races (“your mother was a… STOP! …hamster… STOP!,STOP!,STOP DAMMIT! …and your father … PLEASE STOP!..smelt of elderberries… it sounds like you’re asking for some information about elderberries. Can you tell me more about what you’d like to know about elderberries?.. STOP!”).
It feels like we need the opposite of a wake word (I’ll call it a “sleep word”) which starts listening for a specific keyword AFTER the Voice hears the wake word and will abort all activity if it hears that sleep word. I realize that will require a bunch of additional processing but it would be useful for people who have the horsepower (and who are also too lazy to get up from their couch).