Quranic-Maqam-Classification-using-Deep-Learning

Table of Contents

1. Project Overview
2. Demo
3. Features
4. Technologies Used
5. Maqam Classes
6. Web Interface Link
7. Model Training and Code
8. Setup Instructions
   • Prerequisites
   • Backend Setup (Flask API)
9. Usage
10. Acknowledgments

Project Overview

This project aims to classify Quranic maqams using deep learning techniques. A Deep Neural Network (DNN) model was trained on the Maqam-478 Dataset, consisting of various Quranic recitations, to automatically predict the maqam of a given audio file. The system utilizes Mel-frequency cepstral coefficients (MFCC) for audio feature extraction.

Additionally, the project includes a web interface to allow users to upload Quranic recitations and receive the predicted maqam.

Demo

https://github.com/user-attachments/assets/8d1d7095-03cb-43d6-b1d2-4facc9db2925

Features

• Automatic Quranic Maqam Classification: Upload an audio file and receive the predicted maqam.
• High Accuracy Classification: The system identifies the maqam with high accuracy and displays the accuracy percentage for each classification. Interactive Confidence Scores: After processing, the model presents the top maqam predictions alongside their confidence scores, allowing users to see how confident the model is for each maqam.
• Deep Learning Model: A pre-trained DNN model is deployed to classify the maqams.

Technologies Used

• Python (Backend, Flask)
• TensorFlow/Keras (Model Training)
• Librosa (Audio feature extraction)
• Google Colab (Model Training)
• HTML/CSS/JavaScript (Frontend for web interface)
• Render (Backend hosting)

Maqam Classes

The following Quranic maqams are classified by the model:

• Ajam = عجم
• Bayat = بيات
• Hijaz = حجاز
• Kurd = كرد
• Nahawand = نهاوند
• Rast = رست
• Saba = صبا
• Seka = سيكا

Web Interface Link

You can access the deployed web interface here: Quranic Maqam Classification Web Interface

Model Training and Code

You can find the model training code in this Google Colab notebook:

• Google Colab Code
• Google Colab Code2

Setup Instructions

Prerequisites

• Python 3.x
• Flask
• TensorFlow/Keras
• Librosa
• Numpy

Backend Setup (Flask API)

1. Clone the repository:

   git clone https://github.com/FatimaALzahrani/Quranic-Maqam-Classification-using-Deep-Learning.git
   
   

2. Install dependencies:

    pip install -r requirements.txt
   
   

3. Run the Flask server:

   ````bash python app.py

The Flask application will start at http://127.0.0.1:5000.

Usage

• Visit the web interface and upload an audio file in .wav format.
• Wait for the rotating progress bar to complete while the backend processes your file.
• The predicted maqam will be displayed in both English and Arabic.

Acknowledgments

• Maqam-478 Dataset: Link to dataset
• This project was developed by Fatimah Mohammed Al-Zahrani.