>>47609652It's more patching different open source components together than programming everything yourself from the ground up. You'll have to program some things yourself but most of it will be editing config files.
Basically speech to text -> LLM -> TTS -> sync with live2D animations.
If you want to cut corners on training the LLM you can also inject text into the LLM output before the TTS to make it say things like heart and wink randomly, or like pladis_dev Aiko to meowify her speech.
Getting comfortable with linux is probably more valuable than grinding a programming language if you just want to make your own AI waifu.