What were you expecting? Using Expectancy Features to Predict Expressive Performances of Classical Piano Music

Abstract

In this paper we present preliminary work examining the relationship between the formation of expectations and the realization of musical performances, paying particular attention to expressive tempo and dynamics. To compute features that reflect what a listener is expecting to hear, we employ a computational model of auditory expectation called the Information Dynamics of Music model (IDyOM). We then explore how well these expectancy features -when combined with score descriptors using the Basis-Function modeling approach -can predict expressive tempo and dynamics in a dataset of Mozart piano sonata performances. Our results suggest that using expectancy features significantly improves the predictions for tempo.

Research in Computational Expressive Music Performance and Popular Music Production: A Potential Field of Application?

“…In these last contributions, instead of using a simple weighted linear combination of the basis functions (LBM-linear basis model), feed-forward neural networks (FFNNs), bidirectional recurrent neural networks (RNNs-see [74]), a combination of FFNN and RNN, and a long short-term memory network (LSTM) were tried on piano solo and symphonic repertoires, showing better prediction accuracy than the original LBM. Another development of the model, therefore, took place with the addition of the relationship between the formation of musical expectations and the corresponding musical performances to the analysis of score features based on basis functions, with significantly positive results [75].…”

Research in Computational Expressive Music Performance and Popular Music Production: A Potential Field of Application?

“…In these last contributions, instead of using a simple weighted linear combination of the basis functions (LBM -Linear Basis Model), Feed Forward Neural Networks (FFNNs), bidirectional Recurrent Neural Networks (RNNs -see [69]), a combination of FFNN and RNN, and a Long Short Term Memory network are tried on piano solo and symphonic repertoires, showing better prediction accuracy than the original LBM. Another development of the model therefore took place with the addition of the relationship between the formation of musical expectations and the corresponding musical performances to the analysis of score features based on basis functions, with significantly positive results [70]. The KTH rule system [43] provides articulation rules, and in particular it takes into account the markings of legato and staccato present in the score.…”