Home Creators Posts Import Register Favorites Logout

Content

💡 Ever wanted to create your own AI character—one that talks, thinks, and has a real personality? In this tutorial, I’ll walk you through setting up the foundation using my open-source GitHub project: project_riko.

🔧 This is the exact codebase I use to power Riko, my cheeky anime co-host with memory, voice, and personality. By the end of this video, you'll have your own AI character up and running, ready to talk and be customized.

✅ What We’ll Cover:

How to clone and run the project_riko repo

How the core modules work: memory, voice, personality loop

Tips for customizing your AI’s behavior and tone

🧠 You don’t need to be an AI expert to follow along—just a basic understanding of Python is enough.

👾 Ideal for:

Creators who want an AI VTuber or digital sidekick

Devs building custom AI NPCs, roleplay bots, or companions

Fans of Riko who want to fork and remix their own version

Files

Comments

BrainFartArt

would be nice to run it with LM studios local server

Silky Chai

I followed this tutorial today and was messing around with your code to make it work with a local llm using oobabooga. It was surprisingly easy, your code is so simple, readable and easy to handle. Congrats! You earned my membership! (as soon as I get paid :P). Congrats! You're amazing!

BrainFartArt

would you mind sharing the changes, im trying to use it with LM studio

Silky Chai

But of course my dude! So, you can go to my github, my fork from Project Riko, with this link: https://github.com/SilkyChai/riko_project_fork My changes are on the branch luna-tweaks I changed some things that should be of interest to you: - First, I'm using dotenv, so you gotta create you .env file and fill with your environment variables. I included a .env.example file to see the env vars I added. - You need Oobabooga to test this. You can easily google how to do it. - In Oobabooga, be sure to run it with the api enabled. You can also enable it in the GUI, Session tab, mark the checkbox "api", the restart button and then go to the Model tab and load your model. - My changes are done for Linux, which should not impact you, as most were on the SoVITS files. Bear this in mind, however, in case I did any linux related change to the Riko Project folder and forgor. - I have added a gitignore file as well, so if there's something missing and you can't regenerate, you can refer to the og repo by Rayen. Also, please let me know, as this is not intended. - Finally, don't forget to credit Rayen for giving us this amazing tool! ** Disclaimer ** This is my first project doing AI stuff, and also my first project in Python. I am a C# dev, so excuse me if I messed up or don't follow python best practices, these are my baby steps in the AI world :P Let me know if you fixed it, and consider doing a fork and sharing your changes to enhance Project Riko!

BrainFartArt

oh wow big thx, i will try setting up ooga when im back from my work trip, my SoVITS is crashing at 10% on the TTS_test - i guess i have to try it on linux or reinstall since the FAISS gpu is not compatible with a 5090(?) for now, i tried the CPU version but that might be why its crashing on "Decoding EOS"

Silky Chai

Oh, I hope you can make it work. Yeah, I'm not sure about the latest nvidia architecture but it has been around for a year, right? So I find it weird that it doesn't have support. I only have a 4070 ti Super. I've been considering putting together another pc just to run LLMs as my career seems to be evolving that way xD. Wish you the best and let me know if you need anything. I don't promise to be able to help you, as I told you this is my first time with these technologies, but I can surely help you with any problem with my own code.

BrainFartArt

the ooga part seems to be working, i can see api calls and read (intelligent) replies - just the sovits is still buggy - i have set it uo in wsl but i have to go to work now :/

RIP930

Thanks I followed your tutorial. I successfully modify your script so it work with my llm (Ollama) witch works perfectly fine. But you were write she's kinda mean xD

AceTasks

Drop the detailed video on how to fine tune PLZZZZZZZZZZZZZZ

ismail Ouad

i can't run the go-api file of gpt sovits i don't have a GPU and when i modified the code so it works with the CPU i still get an error

Horizon Variety Show

Does this also require OpenAI credits? I keep getting an issue saying that I don't have enough

Kaiger NZ

yeah, its really cheap tho 5 bucks will last a while

Rhinos Designs

hello, how to modify or change/ train another voice, i have got samples to train, do you have tutorials for it too please ? thank you.

Witherside

Yooo do you have some sort of fork or explanation to how you did this? would be a lifesaver. thanks.

Dylan Trisciuzzi

I know this is from 3 months ago, but did you end up finding anything out on how to do this?

Nehal

HEY i am trying to do similar thing but with groq ,,

Willvern

Can Use Another API, like Google Gemini?

Betrayer

did u find a answer