anuran-roy / nlp-project

Offline speech recognition-enabled assistant

This is a simple offline voice recognition-based assistant purely built on Python3. It uses Vosk as the backend. A wrapper s2t_class.py has been implemented from scratch that recognizes text from speech using Vosk APIs. The main app excelmod.py contains the functions required to make the voice commands work. The file config.yaml maps the commands to their respective functions as a collection of key-value pairs. The corresponding keywords then trigger their respective functions.

3rd party modules used

1. Vosk API by Alphacephei
2. Vosk small American English model
3. OpenPyXL

How to use

1. Clone the repository.
2. (Optional but recommended) Create a virtual environment using venv or virtualenv.
3. Install dependencies using pip install -r requirements.txt
4. Run python excelmod.py
5. Enjoy!