An interactive AI character with voice input, voice output, and profile image generation—all running locally with Nexa SDK and powered by Llama3 Uncensored Model. Enjoy a private and immersive experience, right on your device
This project is an interactive AI character with voice input, voice output, and profile image generation—all running locally with Nexa SDK and powered by Llama3 Uncensored Model. It offers two options for voice output: using the Bark model for on-device text-to-speech or the OpenAI TTS API for cloud-based text-to-speech. Bark will be slow to generate speech without using GPU, but it's on device. The OpenAI TTS API has the advantage in terms of speed, but it is cloud-based and requires you to have an OPENAI API KEY. Each option is designed to provide flexibility based on the user's resources and preferences.You can also choose other options according to your preference.
Key features:
pip install -r bark_requirements.txt
streamlit run bark_voice_out/app.py
pip install -r openai_requirements.txt
Set your openai api key in your terminal. If you don't have a key, ignore this step:
For Windows:
set OPENAI_API_KEY="your_api_key"
For macOS/Linux:
export OPENAI_API_KEY="your_api_key"
Set voice output on in your terminal. If you don't set openai api key or you don't want to use voice output, leave it false:
For Windows:
set VOICEOUT=true
For macOS/Linux:
export VOICEOUT=true
Run the Streamlit app:
streamlit run openai_voice_out/app.py
Start a chat with text or voice as you like
bark_voice_out/app.py
: main Streamlit app using Bark for voice output
bark_voice_out/utils/initialize.py
: initializes chat and load model
bark_voice_out/utils/gen_avatar.py
: generates avatar for AI Soulmate
bark_voice_out/utils/transcribe.py
: handles voice input to text transcription
bark_voice_out/utils/gen_response.py
: handles text and voice output
openai_voice_out/app.py
: main Streamlit app using OpenAI TTS API for voice output
openai_voice_out/utils/initialize.py
: initializes chat and load model
openai_voice_out/utils/gen_avatar.py
: generates avatar for AI Soulmate
openai_voice_out/utils/transcribe.py
: handles voice input to text transcription
openai_voice_out/utils/gen_response.py
: handles text and voice output
More new features and improvements will follow: