/jp/ - Otaku Culture

>>25150592
Because this product is 80% hardware, 15% voice acting/models/animations and 5% is ELSE IF ELSE IF ELSE IF
If you know a bit about these stuff, you can easily make a LCD screen in a tube, get some MMD model, spend months animating and trying to make Vocaloid not make your ears bleed when it talks. Then add some firebase to imitate sms notifications on your phone and, some cloud or local solution for voice recognition you are pretty much set. The problem is that even after all of this, your Miku will only just answer to bunch hardcoded commands and adding new stuff requires you to animate and synthesize dialogues again and again.

That would be best what you can do, unless you somehow find thousands of hours of learning materials of "how to behave like a proper waifu" and teach some GPT-like ml model. Maybe she wouldn't be able to think for herself or do anything too amazing, but at least will generate something suitable as a response.