This is a work code for the coursework "Application of Topological Data Analysis in Music Information Retrieval" in terms of Master Program "Data Science" in the first year of study.
This code uses as base dataset covers80 dataset.
Here is a small description of each file:
- In this .py file I create point clouds from .mp3 files from dataset
- In this .ipynb notebook I place all data (point clouds) in DataFrames in order to access them easily.
- In this .ipynb notebook I create from point clouds persistence diagrams.
- In this .ipynb notebook I make Rips-filtration on point clouds and extract persistence diagrams from them, then extract topological features from persistence diagrams.
- In this .ipynb notebook I create paired dataset from feature datasets and split it test/train, with new mutual features.
- In this .ipynb notebook I evaluate machine learning models on extracted features of mutual datasets.
- In this .ipynb notebook I examine another approach for cover detection task, which is using Siamese Neural Network.