Stabilizing Training with Soft Dynamic Time Warping: A Case Study for Pitch Class Estimation with Weakly Aligned Targets

Code base for reproducing the results presented in

J. Zeitler, S. Deniffel, M. Krause, and M. Müller. "Stabilizing Training with Soft Dynamic Time Warping: A Case Study for Pitch Class Estimation with Weakly Aligned Targets", in Proc. of the 24th Int. Society for Music Information Retrieval Conf., Milan, Italy, 2023.

© Johannes Zeitler ([email protected]), 2023

Overview

• ./Code/: contains all code for running the experiments

• ./Data/: Location for Schubert Winterreise Dataset and computed HCQTs

• ./Logs/: Training log files

• ./Models/: All trained models from the paper

Getting started

• download the Schubert Winterreise Dataset (SWD) from Zenodo and store it in ./Data/
• install the conda environment (conda env create -f environment.yml)
• run ./Code/01_Preprocessing.ipynb to compute HCQTs for the SWD recordings
• run ./Code/02_Training.ipynb to train deep chroma estimators with SDTW loss
• run ./Code/03_Evaluation.ipynb to compute F-measures of trained models (= reproduce the results from Table 2 in the paper)
• run ./Code/04_SoftAlignmentMatrix.ipynb to compute the soft alignment matrices (= reproduce Figure 4 in the paper)

Acknowledgements

This work was supported by the German Research Foundation (DFG MU 2686/7-2). The authors are with the International Audio Laboratories Erlangen, a joint institution of the Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) and Fraunhofer Institute for Integrated Circuits IIS.

This repository builts upon code from christofw/multipitch_mctc and Maghoumi/pytorch-softdtw-cuda.