Per Picovoice, Porcupine is able to detect multiple wake words simultaneously (with no additional CPU load). Can we make that capability also available in Rhasspy (when using Porcupine)?
The typical use case I am after is the following:
I want to use two wake words: “Smarty-pant” and “Hey dumb dumb”.
When “Smarty-pant” is detected, I can directly dialogue with Chat GPT (=> no expectation of privacy).
When “Hey dumb dumb” is detected, everything stay on device (=> privacy enforced)
Examples of dialogues:
Me: Hey Smarty-pant
GPT: What is your question?
Me: Tell me what is the height of the eiffel tower
GPT: The height of the eiffel tower is … blah blah
Me: Hey dumb dumb,
Rhasspy: beep
Me: Switch on the light
Rhasspy: confirmation beep
Right now, I have implemented this differently using a single wake word and introducing an intermediate step before talking to GPT.
Me: Computer
Rhasspy: beep
Me: I have a question
GPT: What is your question?
etc
This is inconvenient because it takes more time to ask my question to GPT.