Pitch detection algorithms have not been extensively explored for non-eurogenetic musical traditions, despite the rich theoretical foundation provided by ethnomusicology. In an effort to address this gap, our research takes a practical approach by examining contemporary pitch detec- tion methods and applying them to a case study. We propose a pitch detection tool for gamelan gong kebyar, a traditional Balinese ensemble. By focusing on this particular case study, we hope to provide insights that can be applied to develop tools for other music traditions.
We used Python 3.11.7. You can run pip install -r requirements.txt to install the dependencies. We recommend using a virtual environment to avoid conflicts with other projects.
pipeline.ipynb: This notebook implements a pipeline described in a report. It takes an audio file as input, extracts a gamelan tuning vector, compares it with three different theory-based tuning vectors, and predicts the regency of the gamelan.
experiments.ipynb: This notebook contains several attempts that haven't produced promising results and have thus not been included in the main pipeline.
helpers.py: This file contains helpers functions used in the implementation of the main pipeline (e.g. find_stable_regions, find_scale).
gamelan_tuning_analysis.ipynb: This notebook contains the analysis and visualization of the distribution of the data in the Toth spreasheets. It also extracts the pemade tunings and stores them in a pickle file.