FAQ

Frequently Asked Questions

---

Speech Recognition and Word files

• How do you add your own "hotword" or "wake word"?

You go to the kitt.ai website, and open an account, then record your own words. After you have recorded the words you can download them in the *.pmdl format. OR You can add and download other, already created words in the universal *.umdl format.

• How do you change a voice recognition wake word?

Just copy your word model file to what is crrently used.

• How do you convert a *.pmdl to *.umdl word model?

You don't, you simply change the file name. What you need to understand is that these files are models for the speech recognition to match. So the only difference is how the model was trained. The personalized model as indicated by the p in *.pmdl, is trained only by you, and by repeating the same word a very few times (often 3). The universal model, as indicated by the u in *.umdl, is trained by several other people, often in the hundreds.

Weird Responses from: Google Assistant

• Why does GA respond with: "Give me permission to help you."?

For example, you test by asking GA a simple question like:
Request: "Snowboy! -- What time is it?"
Response: "Give me permission to help you."

Because you have to tell the GA what it is allowed to do. So you have to enable at least the following options in your Activity Controls:

1. Web & App Activity (including all Chrome browsing and activity)
2. Device Information
3. Voice & Audio Activity

---

ALSA configuration

There are (really) only two choices for setting up voice interaction with a Raspberry Pi. This is because there is no Microphone input on the circuit-board of the RPi's, so you always need to add an external USB sound-card that has a MIC input. The output (AUX) can be either through that same USB sound-card or HDMI or the on-board 3.5mm Audio JACK. The audio jack is crap, so use the USB card! So your options are:

1. Use the USB sound-card for both input and output.
2. Use the USB sound-card for input and the HDMI for the output.

🔴 We strongly recommend to use option 1 !

So this simplifies your ALSA audio setup a whole lot!

Assuming your card is in "slot" 1. Your ~/.asoundrc should look like this, unless you have made or are using other modifications.

pcm.!default {
  type asym
   playback.pcm {
     type plug
     slave.pcm "hw:1,0"
   }
   capture.pcm {
     type plug
     slave.pcm "hw:1,0"
   }
}

ctl.!default {
        type hw
        card 1
}

Installation