Description

Voice assistant made with python 3 for Windows. It uses OpenAI's Whisper to convert speech to text. Speech Recognition is offline but online commands can be added. It understands English only.

You can get the latest build from the releases section.

Development Environment Setup

1. Install Latest Version of Python 3 (Preferably 3.9.9 but should work on 3.8-3.10) and add to path
2. Open terminal in the project directory.

• If you have a GPU that supports CUDA toolkit then the performance can be increased.
• Remove the # from the last line in the requirements.txt file.
• NOTE: This will increase the download size by approx 2.0 GB.

3. Create a virtual environment (Optional but Recommended)
   1. Type python -m venv env.
   2. Type .\env\Scripts\Activate.ps1
   • Note : On Windows, it may be required to enable the Activate.ps1 script by setting the execution policy for the user. You can do this by issuing the following command: Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope CurrentUser
4. Type pip install -r requirements.txt
5. "assistant.py" is the main file.

• The program will download the "base.en" whisper model on the first launch (Approx. 140 MB).

Build

1. Install pyinstaller
2. Open terminal in the project directory.
3. Type pyinstaller .\Sentinel.spec
4. Application is built in the dist folder.

Commands

Open urls

1. Open youtube
2. Open stack
3. Open udemy
4. Open gmail
5. Open whatsapp

Launch apps

1. Open chrome (You may need to edit path in recognizer.py)
2. Open code (You may need to edit path in recognizer.py)
3. Open anaconda (You may need to edit path in recognizer.py)

Search

1. Youtube "anything you want" (opens in default browser)
2. Google "anything you want" (opens in default browser)

Shortcut to start listening for command-

"l_ctrl + l_alt + l_shift" (starts listening for the voice command) "l_ctrl + l_alt + q" (Closes the app) "l_ctrl + l_alt + c" (Calibrate the microphone) "l_ctrl + l_alt + s" (Toggle the window)