AI Voice Assistant with Personality

This project is an AI-powered voice assistant that allows you to run commands, open apps, or ask for information—all while interacting with a personality!

Requirements

Before you get started, make sure you have the following:

1. mpg123 (for audio playback):

   • On Rhel, Fedora based distros:

     sudo dnf install mpg123

   • On Debian based distros:

     sudo apt update
     sudo apt install mpg123

   • Using homebrew (for Macos):

     brew install mpg123

2. Ollama (for natural language processing):

   • You’ll need this dependency installed to handle advanced interactions. Make sure to follow the installation instructions on Ollama’s official site for your platform.

3. Vosk Language Models (for offline speech recognition):

   • Download a Vosk language model and place it in the models/ directory in the project root. Make sure to update const li = new Listener(timeout, model name, activation word); function in index.ts.
   • For instructions on downloading models, visit the Vosk models page.

4. Internet Connection (for Text-to-Speech):

   • TTS functionality requires an internet connection (I haven't found a satisfactory offline TTS, so this will have to do for now).

Installation

1. Clone the repository:

   git clone git@github.com:omarellahi/vox-ai-assistant.git
   cd vox-ai-assistant

2. Install required dependencies:

   • Make sure you have mpg123 installed (as shown above).
   • Install Ollama according to the official documentation.

3. Download the Vosk language model and place it in the models/ directory.

Usage

• To run the voice assistant:

  npm build //run this to transpile ts to js
  node .

• The assistant will wait for your voice commands. You can ask it to run commands, open apps, or give you information.

Notes

• The assistant’s personality adds a fun twist to your interactions.
• Make sure the mpg123 package is installed to play audio responses.
• Ollama is necessary for processing advanced commands, so ensure it is properly set up.
• Vosk works offline for speech recognition, but TTS (Text-to-Speech) requires an internet connection.
• Requires you to create history.context file in the root of the project.

Enjoy your new AI voice assistant with a personality! As of now it has only been tested on my personal machine with Fedora. So please feel free to open an Issue or create PR for feature or bug fixes. I would appreciate it!