Excalibur

A simple voice assistant that respects your privacy. All voice assistants and smart speakers use proprietary systems and are spying on us 24/7. The goal of this project is to replace all of the spyware voice assistants in your digital life.

Features

🟥 Confirmed not working yet

🟨 Partially working/in development

🟩 Fully working

🟦 Needs testing

⬛ Not implemented

Feature	Windows	MacOS	Linux
Speech recognition	🟦	🟦	🟩
Text to speech	🟦	🟦	🟩
Basic questions	🟦	🟩	🟩
Advanced questions*	🟦	🟦	🟩
Time	🟦	🟩	🟩
Weather	🟦	🟩	🟩
Smart device control	⬛	⬛	🟨
Music control	⬛	⬛	⬛
Nextcloud integration	⬛	⬛	⬛

* Advanced questions are questions that require searching on the internet

Deployment

Currently it is meant to be deployed on a desktop or laptop computer. In theory you can deploy it on a small computer like a raspberry pi and replace a smart speaker.

Requirements

* Optional

** Optional but recommended

*** Optional but highly recommended

• Ollama
• Vosk model
• Internet connection
• IP to airport server
• Fast graphics card***

Instructions

* Optional

** Advanced users only

1. Install Ollama
2. Download a Vosk model
3. Put the location of the Vosk model in the vosk_model_path in config.py
4. Set a custom wake word by changing wake_word in main.py*
5. Set a custom LLM model by changing ai_model in main.py**
6. Run ollama serve in a terminal window
7. Run the IP to airport server in a terminal window
8. Connect a microphone and speaker to your computer (you can use a built-in one)
9. Install the required python dependencies listed in the top of main.py
10. Follow the instructions in personalization TEMPLATE.txt
11. Enter the required API keys in api_keys TEMPLATE.py then rename it to api_keys.py
12. Customize config.py
13. Run python main.py in a new terminal window
14. Give microphone access to the program if necessary