passive listen for wake word and active listen after

[Carmine Marino]

Hope this makes sense.  I have jasper working great at the moment.  I am using wit.ai for stt.   I thought when jasper listened for the wake word it did it locally then after the wake word it listens and passes audio to wit.ai.  This doesn't seems to be the case.    When I start Jasper it sends all audio to wit.ai.   

Is it possible to listen for the wake word using pocket sphinx and once the wake word is heard use wit.ai ?  If so how do I set this up?

Thank you in advance.

Carmine

[Carmine Marino]

I figured it out.  I had to look through the code.  I added the stt_passive_engine setting to the config file.   I could see in the code that if this stt_passive_engine setting isn't in place it just uses the stt_engine setting which in my case is wit.ai.

Here is an example of my profile.yml for anyone else looking for this answer. 

carrier: ''

first_name: John

gmail_password: ''

last_name: Doe

phone_number: ''

prefers_email: true

timezone: America/New_York

location: New York

latitude: 40.099762244898

longitude: -74.0442383265306

darkSkyKey: 17be4b36614e8944907c17927695a095

locationIQKey: beb9b500f8a8f1

stt_passive_engine: sphinx

pocketsphinx:

    hmm_dir: '/usr/share/pocketsphinx/model/en-us/en-us'

    fst_model: '/home/pi/phonetisaurus/g014b2b.fst'

stt_engine: witai-stt

#stt_engine: witai

witai-stt:

    access_token: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

tts_engine: pico-tts