This repository contains the code used to generate the symbolic data analysis performed on the SymbTr dataset used for the research project Temporal Evolution of Makam and Usul Relationship in Turkish Makam.
Turkish makam music is mainly rooted in an oral tradition taught in a master-apprentice setting where the pieces are learned through repetition. Its monophonic compositions feature a melodic structure called makam and rhythmic patterns called usul. Since Turkish makam features a larger number of tones per octave compared to Western music as well as a multitude of idiosyncratic rhythmic structures, analysis models that work under the structural premises of the latter are oftentimes not directly applicable to makam. The melodic progression, seyir, is considered an important feature to describe a makam. Previous works on computational analysis of Turkish makam utilize methods such as pitch histograms, n-gram analysis and short-time pitch histograms and are therefore based on the melodic content of a piece or analyze rhythm separately. This project explores the descriptive potential of combined temporal analysis of rhythmic pattern and melodic progression together using a subset of SymbTr, a symbolic dataset that contains 2200 scores with 155 different makams (pl. makamlar) and 88 usuls (pl. usuller).
This repository can be used for replicating the results presented in the paper by using the attached jupyter notebook. An explanation can be found in each cell in order to understand how to run them appropriately. A copy of the dataset is provided along with a script that downloads and uncompresses it directly from the authors' repository release.
Depending on your setup, some dependencies may be needed. There are collected in the requirements.txt file and can be installed by running the following command:
pip install -r requirements.txtIn order to use the jupyter notebook provided in the code folder, it is necessary to have a working copy of the SymbTr-2.4.3 dataset. One is already provided in the code/dataset folder, otherwise it can be downloaded by running the download_dataset.py script located in the root folder of the repository.
python download_dataset.pyThe config.ini contains some paths that are used in the provided jupyter notebooks. notation_app_path is the location of the notation software of choice in your computer that is used to display scores. Muse Score is highly recommended as it is free and cross-platform. Additionally, music_xml_path is the path to the MusicXML files used for the analysis. By default it is stored inside the dataset's folder in dataset/SymbTr-2.4.3/MusicXML but you can adapt it to your own needs.
Finally, dataset_explorer.ipynb notebook in the code folder contains all the necessary cells to generate the analysis and plot used in the paper directly from the dataset files. Some cells may take several minutes to complete and for this reason, precomputed plots are provided in the precomputed_plots folder.
The steps 1 - 4 should contain all the necessary information to explore the data, but here is a quick summary of the folder structure:
code: This folder contains the necessary notebooks and custom packages to generate the plots shown in the paper.config.ini: Configuration file with some necessary variables described in (3).dataset: Folder containing the SymbTr-2.4.3 dataset.dataset_explorer.ipynb: Main notebook used to analyse data and generate the plots presented in the paper.packages: A series of classes defined to manipulate the dataset content. Additionally, some JSON and .musicxml files that contain makam and usul definitions are provided.
download_dataset.py: Script to download the SymbTr-2.4.3 dataset and put it in the default location.paper: Resulting paper (to be added after peer reviewing).precomputed_plots: Folder containing pre-computed plots shown in the paper. The naming convention of each file is explained indataset_explorer.ipynb.