basismixer

The Basis Mixer is an implementation of the Basis Function Modeling framework for musical expression.

The basic idea in this framework is that structural properties of a musical piece (given as a score), which are believed to be relevant for performance decisions, can be modeled in a simple and uniform way via so-called basis functions: numeric features that capture specific aspects of a musical note and its surroundings. A predictive model of performance can then predict appropriate patterns for expressive performance dimensions such as tempo, timing, dynamics and articulation from these basis functions.

Setup

The easiest way to install the basismixer package is with pip.

git clone https://github.com/OFAI/basismixer.git
pip install ./basismixer

Optionally, we include a Conda environment for setting all of the dependencies of the package.

git clone https://github.com/OFAI/basismixer.git
cd basismixer
conda env create -f environment.yml

Usage

Here is a quick guide for using this package.

Tutorial

We prepared a collection of Jupyter notebooks demonstrating this package, which can be found here. These notebooks were presented during the tutorial Computational Modeling of Musical Expression: Perspectives, Datasets, Analysis and Generation presented at the ISMIR 2019 conference in Delft, the Netherlands. You can test these notebooks online.

Generating an Expressive Performance

If the package was set up using the provided conda environment, remember to first activate the environment:

conda activate basismixer

To generate an expressive performance

cd path/to/basismixer
./bin/render_piece [OPTIONS] score_file midi_file.mid

The score (score_file) can be in one of the formats supported by the partitura package.

This repository includes a small model trained on the Vienna 4x22 dataset. This model was prepared for the tutorial on Computational Models of Expressive Performance presented at the ISMIR 2019. This is a demonstration model and is not intended to represent the state-of-the-art.

For more help see

./bin/render_piece -h

Training a Model

See notebook 03_predictive_models.ipynb in the tutorial.

Citing this work

• C. E. Cancino-Chacón and M. Grachten (2016) The Basis Mixer: A Computational Romantic Pianist (2016) In Late Breaking/Demo at the 17th International Society for Music Information Retrieval Conference (ISMIR 2016). New York, New York, USA (pdf)

   @inproceedings{CancinoChacon:2016wi,
   	Address = {New York, NY, USA},
   	Author = {Cancino-Chac{\'o}n, Carlos Eduardo and Grachten, Maarten},
   	Booktitle = {Proceedings of the Late Breaking/ Demo Session, 17th International Society for Music Information Retrieval Conference (ISMIR 2016)},
   	Title = {{The Basis Mixer: A Computational Romantic Pianist}},
   	Year = {2016}}

• C. E. Cancino-Chacón, T. Gadermaier, G. Widmer and M. Grachten (2017) An Evaluation of Linear and Non-linear Models of Expressive Dynamics in Classical Piano and Symphonic Music. Machine Learning 109, 887–909. (link)

   @article{CancinoChaconEtAl:2017,
   	Author = {Cancino-Chac{\'o}n, Carlos Eduardo and Gadermaier, Thassilo and Widmer, Gerhard and Grachten, Maarten},
   	Journal = {Machine Learning},
   	Number = {6},
   	Pages = {887--909},
   	Title = {{An Evaluation of Linear and Non-linear Models of Expressive Dynamics in Classical Piano and Symphonic Music}},
   	Volume = {106},
   	doi = {10.1007/s10994-017-5631-y},
   	Year = {2017}}

• C. E. Cancino-Chacón (2018) Computational Modeling of Expressive Music Performance with Linear and Non-linear Basis Function Models. PhD thesis. Johannes Kepler University Linz, Austria (pdf)

   @phdthesis{Cancino2018thesis,
   	Address = {Austria},
   	Author = {Cancino-Chac\'on, Carlos Eduardo},
   	Month = {December},
   	School = {Johannes Kepler University Linz},
   	Title = {{Computational Modeling of Expressive Music Performance with Linear and Non-linear Basis Function Models}},
   	Year = {2018}}

License

The code in this package is licensed under the GNU General Public License v3 (GPLv3). For details, please see the LICENSE file.

The data and sample trained models included in this repository (e.g., the models in basismixer/assets/sample_models) are licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0.

Data and model files usually have (but are not limited to) .npy, .npz, .h5, .hdf5, .pkl, .pth or .mat file extensions.

If you want to include any of these files (or a variation or modification thereof) or technology which utilizes them in a commercial product, please contact Gerhard Widmer.

Acknowledgements

This research has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme under grant agreement No. 670035 (project Con Espressione).