Speech-To-Text Django Web Application

Demonstration

Home Page

The html document of this page also includes a script that establishes a socket connection to the backend and processes input commands. The response of the page html is the change in the states of the button containers.

Commands Recognition Page

This page exists to demonstrate the ability to control web elements in real time and the ability to visualize the input sound wave.

Transformer Speech-To-Text

This page tells about the result of the pre-trained transformer model trained by us.

Ready-made Speech-To-Text model.

We also tried to deploy a library model of speech recognition in our application - Silero-VAD.

Models

for a description of the process of training models, you can go to this repository with Jupyter Notebooks.

https://github.com/Begelit/Speech-To-Text-Training

P.S. Thanks to Kevin(Kaggle - kevinvaishnav) for participating in the project, teaching Transformer model and providing the necessary resources!

https://www.kaggle.com/code/kevinvaishnav/stt-readymade/notebook

Sources

Commands Recognition Model

For successfully use the model that recognizes your commands, you need to download this file and copy it to this directory "./Speech-To-Text-Web-App-Django/Django_Web_Application/commands_model/".

Transformer Model

Fir use transformer model in this web application you need download "pytorch_model.bin" file from this link and copy it to this directory "./Speech-To-Text-Web-App-Django/Django_Web_Application/transformer_model/"