Music-Genre-Classification

Music genre classification using GTZAN dataset, a dataset containing 1000 labeled audio files of 10 different genres.

The model takes as input the Mel Spectrogram of the song and analyzes the image using a Convolutional Neural Network (CNN). The output is a vector containing the score associated to each genre.

Dataset

The dataset contains 1000 labeled audio files (belonging to famous songs) all having a length of 30 seconds of 10 different genres. The genres are:

• Blues
• Classical
• Country
• Disco
• HipHop
• Jazz
• Metal
• Pop
• Raggae
• Rock

The dataset can be downloaded here. Place the content in data/raw/ folder. For this project only the folder genres_original will be used. It contains the audio files in .wav format.

Installation

Place in the current directory. Clone the current conda environment or install package directly from requirements, then, install the project package.

Cloning the code environment

The conda environment has been exported using the command

conda env export > environment.yml

Type the following instruction to clone the environment

conda env create -f environment.yml

Install from requirements

The requirements has been generated using (check this):

pip list --format=freeze > requirements.txt

Install package requirements using pip

pip install -r requirements.txt

or conda

conda create --name <env_name> --file requirements.txt

Install the project package

The project package can be install using

pip install -e .

The project folder genre_classification is installed as a python package and it can be now referenced anywhere.

API service

Once an experiment is executed and a trained model is ready in the experiments folder (follow the guide in genre_classification), the API service can be used. The API service is used to make inference on a new and unlabeled wav file located in /data/external folder.

Start the server by running

uvicorn app:app --reload

Run your browser to test the service API.

Go to the address

http://127.0.0.1:8000

or

http://127.0.0.1:8000/docs

Project Organization

├── README.md          <- The top-level README for developers using this project.
├── data
│   ├── processed      <- The final, canonical data sets for modeling.
│   └── raw            <- The original, immutable data dump.
│   └── external       <- External data to use for inference. Wav files without label
│
├── models             <- Trained and serialized models, model predictions, or model summaries
│
├── notebooks          <- Jupyter notebooks. Naming convention is a number (for ordering),
│   ├── 1.0-exploration.ipynb
│   ├── 2.0-google-colab.ipynb
│   └── 3.0-visualize_embeddings.ipynb
│
├── requirements.txt  
│
├── setup.py           <- makes project pip installable so src can be imported
├── src                <- Source code for use in this project.
│   ├── __init__.py    <- Makes src a Python module
│   │
│   ├── data           <- Scripts to download or generate data
│   │   └── make_dataset.py <- Generate the final dataset used for modelling
│   │
│   ├── models         <- Scripts to train models and then use trained models to make
│   │   │                 predictions
│   │   ├── config.py
│   │   ├── cnn.py
│   │   ├── dataset.py
│   │   ├── train.py
│   │   ├── evaluate.py
│   │   ├── embeddings.py
│   │   └── inference.py
│   │
│   └── visualization  <- Scripts to create exploratory and results oriented visualizations
│       └── visualize.py
│
└── app.py              <- FastAPI service

---

Project based on the cookiecutter data science project template. #cookiecutterdatascience