Audio Classification App

Overview

This is a prototype Django application that allows users to upload audio files and classify them using machine learning techniques. The app uses Principal Component Analysis (PCA) for feature reduction and K-Nearest Neighbors (KNN) for classification. The application is designed to be extendable, allowing for the addition of more data and classes in the future.

Features

Upload Audio Files: Users can upload audio files through a web interface.
Feature Extraction: Extracts Mel-Frequency Cepstral Coefficients (MFCC) from the audio files.
Dimensionality Reduction: Uses PCA to reduce the dimensionality of the extracted features.
Classification: Classifies the audio files using a pre-trained KNN model.
Results Display: Shows the predicted class of the uploaded audio file.

Installation

Clone the Repository:

git clone https://github.com/yourusername/audio_classification.git
cd audio_classification

Create a Virtual Environment:

python -m venv venv
source venv/bin/activate  # On Windows use `venv\Scripts\activate`

Install Dependencies:

pip install -r requirements.txt

Run Migrations:

python manage.py migrate

Start the Development Server:

python manage.py runserver

Access the Application:
Open your web browser and go to http://127.0.0.1:8000/.

Usage

Upload an Audio File:
- Navigate to the upload page.
- Select an audio file and click the "Upload" button.
View Classification Results:
- After uploading, you will be redirected to a results page showing the predicted class of the audio file.

Concepts

Feature Extraction

The application uses the librosa library to extract MFCC features from the audio files. MFCCs are commonly used in audio processing to represent the short-term power spectrum of a sound.

Dimensionality Reduction

PCA is used to reduce the dimensionality of the extracted features. This helps in reducing noise and simplifying the data for classification.

Classification

A pre-trained KNN model is used to classify the audio files. KNN is a simple, yet effective, classification algorithm that assigns a class based on the majority class of the nearest neighbors.

Extending the Application

This application is a prototype and can be extended in several ways:

Add More Data: Collect and label more audio files to improve the accuracy of the model.
Extract Additional Features: Use other audio features like chroma, spectral contrast, etc.
Use Advanced Models: Experiment with more complex models like Convolutional Neural Networks (CNN) or Recurrent Neural Networks (RNN).