This is an AI-assisted extension for EDDI. It is designed to generate more natural-sounding speech not just by using a better TTS engine, but also by using AI to rephrase the text. Also provides interaction capabilities. It aims to be a drop-in replacement for the ingame COVAS, by being an AI assistant.
It even has a small GUI window that can be dropped into VR to see what the assistant is saying.
Notice: This is using an older version of oobabooga's API for LLM interaction, and I have to update it to the current one.
This is before the GUI was made.
I suggest using a virtual environment for this (conda or venv). Also, I recommend using Python 3.9 or higher.
- Clone this repository
- Install the requirements:
pip install -r requirements.txt
- Create a
config.json
file in the root directory of this repository. See config.json.example for an example.
To start the program, just run:
python launch.py