Prop replica modification featuring a Super Inexperienced Person

General Discussion
1

Jumping right in and saying that I know that what I want to do is kind of not the intended use of Mycroft.

I don’t want it to speak to me (ever).

I don’t want it to have much functionality.

I want a system that responds with WAV playback of prerecorded sound files to a handful of recognized keywords. I want to make a functional replica of Chucky, the Good Guy doll from Child’s Play

I need to change the wake word using Precise. I’ve done so with PocketSphinx, but it’s just false positive after false positive (including my wife shutting the door two rooms over). I saw that there was a Skill to change the wake word with Precise, but I’m not understanding the documentation.

It’s been 20 years since I’ve done much of anything on Linux other than installing a few RasPi images for friends and whatnot. I have never programmed in Python, or anything else really.

The full extent of what I want to do:

User: “Hi, I’m (doesn’t matter)” or “Hey, I’m (doesn’t matter)”, “Hello”, or similar.
Picroft: randomly plays one of 3 prerecorded sound files

User: “Sleeping”, “Go to bed”, “Goodnight”, or similar.
Picroft: plays specific sound file

User: “Meaning of life”, “why are we here”, or similar.
Picroft: plays specific sound file

User: “Your new name is Tommy” or similar.
Picroft: plays specific sound file

User: anything rude
Picroft: Plays specific sound file

I can’t imagine this would be terribly difficult for the most part - except for removing the need for a wake word - though I am probably wrong due to my lack of knowledge. I’m not a good written tutorial learner, it took me a few days to get Picroft successfully running on my RP3b+, which probably isn’t a record for slowest attempt, but it can’t be good either.

I guess that leaves me with five main questions:

  1. Can Picroft run with no wake word?
  2. Can I disable TTS without breaking everything?
  3. Can I disable all skills that install by default?
  4. Can someone point me to “The Complete Idiot’s Guide” to making a skill that just plays a sound file and ends?
  5. Can I run the system completely offline on the RP3b+?

Thanks for reading!

2

Hey ShiftySamurai,

Love your clear questions and purpose. Sounds super creepy and awesome :slight_smile:

  1. You could remove all wake words, however you need some way to trigger the listener. There is not currently a way to run continuous speech recognition. Using a remote service for this would be incredibly expensive, and it’s not something we allow. Possibly of more use is to use multiple wake words with Jarbas’s Hive Mind.
  2. The easiest thing would probably be to set the TTS config to something like Mary but with no real service.
  3. To blacklist Skills, you add a list to your configuration, here’s an example line.
  4. Here’s an Audio Output Skill. It has three Adapt intents, one for each type of audio output. You can find more on Skill dev here and intents in particular.
  5. FAQ: Can Mycroft run completely offline?