This repository is a demonstration for a audio classification methodology based on the tonal concept of 'harmonic modes' in Musicology. The concept of 'harmonic modes' exists in various tonal music traditions around the world with distinct alterations. In this project, we apply chroma based supervised mode recognition in the multi-cultural context. The evaluation of algorithms are designed to be performed on the Ottoman-Turkish, Hindustani and Carnatic datasets that were built within the scope of CompMusic Project.

• Please cite the reference below if you use the content of this repository in your work.

[1] Demirel, E., Bozkurt, B., Serra, X. (2018). Automatic Makam Recognition using Chroma Features, in proceedings of Folk Music Analysis 2018 (FMA) https://zenodo.org/record/1239435#.WulmO3VMSfg

In order to use the tools and notebooks, you need to install 'docker' . Docker provides a virtual environment with all the desired dependencies and libraries already installed. In this toolbox for 'Chroma-based Supervised Mode Recognition in Multi-cultural Context', we have used the tools in 'MIR-Toolbox' which contains a set of tools (including numpy,matplotlib, scikit-learn,'Essentia' ) installed and compiled for several Music Information Retrieval applications. For more information regarding toolbox, please refer to https://github.com/MTG/MIR-toolbox-docker :

1. Install docker-compose Follow instructions.

Windows https://docs.docker.com/docker-for-windows/install/

Mac https://docs.docker.com/docker-for-mac/install/

Ubuntu https://docs.docker.com/engine/installation/linux/docker-ce/ubuntu/#install-docker-ce

2. Clone the repository into target directory.

git clone https://github.com/emirdemirel/Supervised_Mode_Recognition.git

3. On MacOS or Windows, run:

docker-compose up

On Linux, run the following (this command ensures that any files you create are owned by your own user):

JUPYTER_USER_ID=$(id -u) docker-compose up

Then accesss http://localhost:8888 with your browser and when asked for a password use mir

4. Open the Jupyter notebook 'DEMO_SuperModeRecog.ipynb' for the DEMO of the work presented in paper [1].

5. In the second step of the DEMO file, you can also apply 'Automatic Mode Recognition' on a single file, using the classifier model explained in the paper, trained on the full dataset.

Emir Demirel emir.demirel@upf.edu