The Basis Mixer: A Computational Model of Expressive Music Performance
The Basis Mixer, opens an external URL in a new window is an implementation of the Basis Function Modeling framework for expressive music performance that was comprehensively described and analysed in
Cancino Chacón, C. (2018). Computational Modeling of Expressive Music Performance Through Linear and Non-linear Basis Function Models, opens an external URL in a new window. Ph.D. Thesis, Inst. of Computational Perception, Johannes Kepler University (JKU), Linz, Austria.
The basic idea in this framework is that structural properties of a musical piece (given as a score),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 (based on deep neural network machine learning models) then predicts appropriate patterns for expressive performance dimensions such as tempo, timing, dynamics and articulation from these basis functions
The Partitura Software Suite
Partitura is an open source Python package for handling richly structured symbolic musical information in a less reductive yet transparent way than the piano roll representation that that is common in MIR. The package allows for representing a variety of note information present in musical scores beyond the onset, duration and MIDI pitch numbers of notes, such as pitch spelling and voicing information. Moreover, it supports musical notions that are not note-related, like measures, and tempo and dynamics markings. Currently, Partitura supports reading and writing MusicXML and MIDI files, as well as reading all formats supported by MuseScore.
In addition to handling symbolic music formats, the package provides some music analysis tools for estimating key signature, pitch spelling, voice information and tonal tension profiles.
The latest version of the software can be downloaded here, opens an external URL in a new window. The full documentation is available online at readthedocs.org, opens an external URL in a new window.
A first version of the tools was presented at the ISMIR 2019 conference:
Grachten, M., Cancino Chacón, and Gadermaier, T. (2019). partitura: a Python Packge for Handling Symbolic Musical Data, opens an external URL in a new window. Late breaking / demo papers, 20th International Society for Music Information Retrieval Conference (ISMIR 2019), Delft, The Netherlands.
The Con Espressione Game Dataset
The Con Espressione Game Dataset, opens an external URL in a new window is a cleaned and documented snapshot [Summer 2020] of the user data (free-text characterisations of perceived expressive performance character) that we collected via the Con Espressione Game, opens an external URL in a new window (see also our main project page). It comprises some 1,500 user descriptions of expressive character relating to 45 performances of 9 excerpts from classical piano pieces, played by different famous pianists. A first preliminary analysis of the data is presented in:
Cancino Chacón, C., Aljanaki, A., Chowdhury, S., Peter, S. and Widmer, G. (2020). On the Characterization of Expressive Performance in Classical Music: First Results of the Con Espressione Game, opens an external URL in a new window. In Proceedings of the 21st International Society for Music Information Retrieval Conference (ISMIR 2020), Montreal, Canada. [Video, opens an external URL in a new window]
The Multimodal Sheet Music (MSMD) Dataset
The Multimodal Sheet Music Dataset (MSMD), opens an external URL in a new window is a synthetic dataset of 497 pieces of (classical) music that contains both audio and score representations of the pieces aligned at a fine-grained level (344,742 pairs of note heads aligned to their audio/MIDI counterpart). It can be used for training and evaluating multi-modal models that enable crossing from one modality to the other, such as retrieving sheet music using recordings or following a performance in the score image. MSMD was first used in the paper:
Dorfer, M., Hajič jr., J., Arzt, A., Frostel, H. and Widmer (2018). Learning Audio-Sheet Music Correspondences for Cross-Modal Retrieval and Piece Identification, opens an external URL in a new window. Transactions of the International Society for Music Information Retrieval, issue 1, 2018.
Tutorial Materials: Computational Models of Expressive Performance
In November 2019, Carlos Cancino, Katerina Kosta, and Maarten Grachten gave a Tutorial on Computational Modeling of Expressive Performance, opens an external URL in a new window at the 20th International Society for Music Information Retrieval Conference (ISMIR 2019), Delft, The Netherlands. Here are the corresponding teaching and experimentation materials:
This project receives funding from the European Research Council (ERC), opens an external URL in a new window under the European Union's Horizon 2020 research and innovation programme under grant agreement No 670035.