DHR2206/Tiny-ML ? reverse-engineered prompt

Build me an offline voice assistant for kids ages 3 to 8 that runs on a Raspberry Pi 5 and never sends audio to the cloud.

It should listen for a wake phrase like “Hey Buddy”, understand simple child friendly commands, and respond with pre made voices. The app should support things like stories, poems, alphabet, phonics, counting, animals, quizzes, jokes, and simple FAQ style answers. I want it to be tolerant of child pronunciation and spelling mistakes, and it should work without keeping recordings.

Please make it so I can install it on the Pi, generate the content once, and then run it locally. Include the basic command line flow, a setup process, and tests so I can check that wake word detection and common commands work. If you need current docs for the speech or voice tools, look them up online first.