Audio-Fingerprinting

Fingerprint-based music and vocals identification app that generates spectrograms, extracts features, applies perceptual hashing, and finds the most similar songs based on fingerprint matching.

Table of Contents

• Prerequisites
• Installation
• Features
• Contributors

Prerequisites

• Python 3.6 or higher

Installation

1. Clone the repository:

   git clone https://github.com/AhmedAmgadElsharkawy/Audio-Fingerprinting.git
   

2. Install The Dependincies:

   pip install -r requirements.txt
   

3. Run The App:

   python main.py
   

Features

• Spectrogram Generation: The program iterates over songs in the shared folder, generating spectrograms for the full song, music, and vocals.

• Feature Extraction: For each spectrogram, key features are extracted, analyzed, and stored in a file. This allows for perceptual hashing to create a unique fingerprint for each song.

• Perceptual Hashing: Extracted features are hashed using perceptual hash functions to generate a compact representation of each song for easier identification.

• Song Similarity Search: Given any input sound file (song, vocals, or music), the program generates a spectrogram, extracts features, and compares it to the Database, outputting a sorted list of similar songs with a similarity index.

• Weighted Average Song Mixing: Users can select two files and control their contribution weight via a slider. The software then treats the combination as a new song and searches for similar songs in the Database, giving preference to the original files based on their similarity index.

Contributors

• AhmedAmgadElsharkawy: GitHub Profile
• AbdullahMahmoudHanafy: GitHub Profile
• MohamadAhmedAli: GitHub Profile
• RawanAhmed444: GitHub Profile