Neuron is a voice assistant created for Ubuntu that interacts with your voice using Vosk, an offline speech recognition API.
Neuron
├── assets
│ ├── data.json
│ └── rep.mp3
├── README.md
├── requierements.txt
├── scripts
│ ├── microphone.py
│ ├── process_commands.py
│ ├── process_speech.py
│ └── utils.py
├── start.py
└── test.py
- Starting & Closing applications
- Maximizing & Minimizing windows
- Writing what you tell him to write
- Managing the display of windows
- Playing musics on youtube
- Looking up requests on google
- Directly opening websites
- Practical voice commands for developpement such as "stop" and "restart"
Due to some python libraries not being supported by docker such as "pywhatkit" I didn't set up a DockerFile. However, I'll try to make this installation easy.
git clone https://github.com/Luzivog/Neuron.git
cd Neuron
sudo apt update
sudo apt install -y libasound-dev portaudio19-dev libportaudio2 libportaudiocpp0
pip install -r requierements.txt
pip install pyaudio==0.2.11 --user
Vosk models are available here, for better results I choosed the "vosk-model-en-us-0.42-gigaspeech" but depending on your machine you may want to consider a lighter model. After downloading the model, move it to Neuron folder and extract it into a folder that you rename "model".
Open a command prompt, make sure you're in the Neuron directory, and run:
python start.py
All the configuration is done in Neuron/assets/data.json You can configure:
- Trigger words: Words that will trigger the assistant.
- Command words: Words that will trigger voice commands such as "type" for the writing action.
- Applications words: Words that refer to specific applications such as "vscode" for Visual Studio Code.
- Website words: Words that refer to specific websites such as "my favorite song" with a youtube link towards it.