Yapster

Convert voice to text and detect gender model

Hosted instance - yap.dumbapps.org

Technologies

• Python based backend

  • Django as the web framework

• Flutter based frontend

File structure

• client - contains the flutter frontend
  • lib - contains the main application code
• ml - contains the notebooks used in training the models
• backend - contains the backend code as well as the trained models
  • model - contains the ASR model
  • voice_model - contains the gender detection model

Building

Using docker (Recommended)

I recommend using docker for building as avoids setting up flutter and python venv shenanigans and avoids installing packages into your system python.

Optionally you can setup Nvidia container toolkit to use your nvidia GPU, but it is not required to build the application.

Build the image

docker build . -t "yapster"

Run the image

docker run -p 8000:8000 yapster 

This will make the website accessible on localhost:8000

Manually building from source

Prerequisites

• You will need flutter installed and accessible, you can check your installation by running

  flutter doctor
  

• Python 3.12 installed

Build the frontend

This will build frontend files in client/build/web/

cd client
flutter build web --base-href=/static/

Build the backend

Install packages

 pip install -r backend/requirements.txt 

Run the backend

By default, it will automatically serve the generated frontend files client/build/web/

py main.py

This will make the website accessible on localhost:8000