🤖 Sparky | PROJECT STATUS: EXPERIMENTAL

Documentation pending!

Sparky is an Open source Voice Assistant, alternatives to Google Assistant and Alexa (oh Siri too btw).

Planning to have integration with Home Assistant which allows the operation of many already existing smart home devices.

How to use

1. Install system dependencies

Windows:

• Python 3.10 or newer

MacOS:

Linux:

• PyAudio
• Python 3.10 or newer (most Linux distros come with python preinstalled)
• Library for espeak: libespeak-dev or libespeak-ng-dev or libespeak1 or libespeak-ng1?
  • Ubuntu: sudo apt install libespeak-ng-dev python3-pyaudio
  • Fedora: sudo dnf install espeak-ng python3-pyaudio.x86_64
  • Arch: sudo pacman -S espeak-ng python-pyaudio

FreeBSD:

On a Virtual Machine

To use sparky on a Virtual machine, you can use Scream to passthrough your mic via a virtual sound card. Here's a video tutorial by Pavol Elsig.

How to run?

# Install python modules
pip3 install -r requirements.txt

# Run sparky
python3 sparky.py

How to develop?

• Get virtualenv for python

pip3 install virtualenv

TODO:

• Documentation, Roadmap
• Rewrite without legacy stuff, using virtualenv
• Maybe use pre-trained models like in this project? Here's a tutorial video.
• Add integration for ChatGPT API and/or Tuxi?
• Build and package via:
  • Flatpak
  • Snap
  • AppImage?
• Implement gTTS alternatives like:
  • Mycroft Mimic 3
  • Festival Speech with Festvox voices v2.4 or v2.1
  • Microsoft SAPI5?
• Option to switch between multiple TTS engines, like switch to Mycroft when offline, and gTTS for when internet is available.