Thank you to @0PT1MUS for the hints provided.
My hints for user :
1º You can use TTS, or record yourself with a mic. There are different online TTS services available, but only one worked for me in the end. I would share the name, but apparently it was considered a spoiler in the first page.
As a hint, there’s one demo available by one company Steve Jobs didn’t like that will help you.
2º Enumerate in order to get additional information about the kind of queries you can perform. Don’t forget to pay attention to the references within that page.
3º There’s one word you need that might be split in two after being parsed by the AI. You can either record the word yourself, or replace the R with “her”.
I like the concept of the box, but I struggled with the TTS.
Let’s see about root.