• Introduction
• Work Process
• Project Features
• Project full Demo
• Project Structure
• How to Run The Project
• Team

Voice Password checker is a web application used for Voice authentication (voice biometrics voice ID or speaker recognition). The system is speech dependent which requires authorized person with correct sentence said. In our appplication, Team Members are only authorized with "Open the door" sentence.

• Voice Feature Extraction depending on MFCCs:

• Models Training for user and sentence identification

• Model Deployment using web technologies (Flask, Html, CSS, JS)

In this web application you can

1. Easily use recorder to check

2. Live Visulaization for input audio features

3. Models Scores of input audio (User and Sentence)

The Web Application is built using:

• Frontend:
  • HTML
  • CSS
  • JavaScript
  • Ajax
• Backend framework:
  • Flask (Python)

The Frontend main function to set the structure of the page and plot the signals and mange the user interface while the backend function is to do on-signal operations like sampling, reconstructing, adding signal, adding noise & removing signal.

main
├─ Notebooks
├─Training sets
├─ static (JS & CSS files)
│  ├─  css
│  ├─  img
│  └─  js
├─ template (HTML files)
├─ app.py (Back-End Server)
├─ Gmm.py (Audio preprocessing and Models Predictions)
├─ utilities.py (Data to draw helpers)
└─ README.md

1. Install Python3 in your computer

Download it from www.python.org/downloads/

2. Install the following packages
   • numpy
   • flask
   • flask_cors
   • Librosa
   • pickle
   • scipy
   • sklearn
   • python_speech_features

• Open Project Terminal & Run

pip install -r requirments.txt

3. Start Server by Running

python app.py

4. Visit http://127.0.0.1:5001

First Semester - Biomedical Digital Signal Processing (SBE3110) class project created by:

• Dr. Tamer Basha & Eng. Abdullah Darwish All rights reserved © 2022 to Team 14 - Systems & Biomedical Engineering, Cairo University (Class 2024)