The project had been used the best TTS (Text To Speech) from Google, the best chat reply AI from ChatGPT GPT 3.5 (until 2023, maybe in future it will have more better ones), Live2D modules, a Live2D character which can be controlled by the VaderSharp2 for Sentiment Intensity Analayze, (removed the character model for sure, since I used paid version, but you can find it on Booth), and also Game Script Bot to play some android games with the help of Zeraniumu Framework.
Yet we get the almost same as a VTuber should looks, but EXTREMELY BORING, no feelings, replying chats like bot (thats what it is, bots!).
Hence, while vieweing VTubers, just enjoy their performance but never belive if they mentioning themselves as "AI", like Kizuna AI, or NeurosamaAI, or even is this Ena AI. Also, for those VTubers out there, do not make society affraid of AIs. PLEASE, AIs are just tools for you to work smarter, easier and reduce your work load, not for fooling your viewers like AI will conquer the world, speak like an ass holes, and pushing all responsibles to "AI"
So from now on, I hope there is no more people finding me and "asking is that VTuber is real AI?" Well, if you want a real AI, here you go, all the source codes and you can just use it for free, MIT license, even commercial usage is permitted. But after you used it, don't blame me while you lose massive of cash for preparing all the stuffs.
- OpenAI API: 1000 token 0.002 USD
- Google TTS: First 1M bytes is free, then 0.16 USD for next future 1M of bytes each month
- Live2D Model: 2000 JPY (About 15 USD)