dicio-android icon indicating copy to clipboard operation
dicio-android copied to clipboard

[FR] verbal dialog mode + STOP word

Open ildar opened this issue 2 years ago • 3 comments
trafficstars

as skills number is growing the time is near when a dialog may be appropriate:

  1. get verbal input
  2. execute
  3. TTS the result outloud
  4. Turn the mic ON for the next command.
  5. If no command given, turn off the mic after timeout

Also very important is: when TTS-ing the result, listen for the user. If he (she) says STOP then stop the TTS.

ildar avatar Jan 30 '23 09:01 ildar

I am not sure I understand how "skill number is growing" is related to "it may be appropriate to turn the mic on for the next command". Turning the mic on automatically is already done when skills ask for more input, e.g. the telephone asking for confirmation.

Adding a simple switch to enable/disable automatic mic turn on, and a simple "stop" skill, should be simple enough.

Stypox avatar Feb 06 '23 17:02 Stypox

Hi!

how "skill number is growing" is related to "it may be

I mean a user may want to ask things from Dicio in sequence. That's what the switch to enable/disable automatic mic turn on is for.

simple "stop" skill

Not quite what I mean. I mean a possibility to stop the current Dicio talking by saying "Stop". I.e. it should work while another skill is still working, not after.

Thanks.

ildar avatar Feb 07 '23 12:02 ildar

I agree with the ideas. My understanding to clarify:

Enhancement A: After one voice command finished, Dicio should listen a little while, if the user wants so interact more.

Example:

  • "Navigation to xyz." -> navigation starts
  • but I want to listen to music on the road so I say "And play some music on Spotify"

Counterpoint: This could be not necessary, when a wakeword is implemented. Maybe a wakeword is even better, because if I talk to someone, dicio may be listening right after the navigation started and interrupt our conversation.

Enhancement B: Stop word Sometimes voice commands get picked up wrong. If so, I don't want to listen to the whole output but try again as soon as possible.

Example:

  • I say "Whats the weather in Paris like today?"
  • vosk picks up "what's the beat way to Paris today?" and starts talking about different routes to paris or different travel modes
  • I want to try again, so I say "stop" -> voice output stops and I can try again. Either via wakeword or dicio starts listening right away.

nils-se avatar May 11 '23 09:05 nils-se