Create the required environment

1. conda env create -f environment.yml

2. conda activate ufo

Directory Structure

The directory contains the following folders:

• data/: contains some audio samples at audio/. The speech chunks get saved at audio/ and MFCC images get saved at images/

• speech_rec_pytorch/: contains files to train a digit recognition model from scratch

• pretrained_model/: contains pretrained model. Replace with the model you want to use. I am using speech_net_aug.pth.tar.

• static/ and templates/: For the Web App part

• misc/: Some utility functions

Files

To run the Web App:

1. export FLASK_APP=application.py

2. flask run.

Then navigate to http://127.0.0.1:5000/.

To run a simple demo

python demo.py -pn <file_name from audio/>

Demo