The purpose of this project was to investigate the potential offered by computational approaches for stylistic characterisation of performances. Chopin's mazurkas were chosen as a varied but well defined repertory that poses interesting performance practice issues and has been frequently recorded; we collected recordings of about 3000 individual mazurka performances, although only a modest proportion was analysed within this project.

Most of the work was based on the extraction of tempo and dynamic data, which was then analysed mathematically, for instance in order to investigate measures of stylistic similarity or to extract musically significant features: this part of the project, attempting to produce meaningful descriptions of performances on the basis of purely objective analysis, was undertaken by Craig Sapp, the project's research fellow, and is best understood in terms of the discipline of Music Information Retrieval (MIR). Nicholas Cook, the project director, explored the potential for applying such analyses within a musicological context, attempting in this way to link empirical data with cultural analysis. The project was so designed as to give rise to a number of analytical tools available to other users.

The project fell into three essentially distinct, though overlapping stages:

  • the development of a range of data capture software. This followed two main paths. The first is based round Sonic Visualiser, the navigation and visualisation environment developed (with some input from CHARM) by the Centre for Digital Music at Queen Mary, University of London, which allows tempo to be manually tracked by means of tapping: Sapp developed a range of plugins for Sonic Visualiser which, among other things, produce refined visualisations enabling more accurate determination of onsets, and capture global dynamic data. The second is the development by Andrew Earis of his Expression Algorithm software, which is optimised for piano music and takes a recording plus a tapped onset profile as its input, generating timing and dynamic information for each note, together with an approximate measure of articulation. Details of these tools may be found in the section of this website on Analysing recordings
  • the development of analytical tools, procedures, and visualisations for working with recordings. The aim of this part of the project was to develop informative, automatically generated representations of certain aspects of recorded performances, particularly of piano music, and to investigate the extent to which such objective analyses can yield meaningful information concerning the complex cultural practices of musical performance and its reception. An unanticipated outcome of this work was the discovery that the Concert Artists recording of the Mazurkas issued in the name of Joyce Hatto was in fact a slightly modified version of the recording by Eugen Indjic; this was the first proof of the infamous Hatto hoax, which became generally known in February 2007. Although some of these analyses involved customised spreadsheets and other approaches developed for individual applications, the project gave rise to a number of general-purpose online tools developed by Sapp and accessible here: they allow you for example to determine global dynamic values at selected points (for instance at each beat), and to generate a variety of correlation plots. Details of these tools may again be found in the section of this website on Analysing recordings
  • an investigation of the musicological potential of these objective analyses. This work, undertaken by Nicholas Cook, focused round four main areas: (i) an investigation of how pianists create the characteristic 'mazurka effect' through timing, dynamics, and articulation, and the different styles of mazurka performance that have resulted; (ii) a study of phrase arching, generally thought to be a definitive aspect of expressive performance but now shown to be a historically and even to some extent geographically delimited practice; (iii) an exploration of how empirical approaches to recordings can be reconciled with interdisciplinary performance studies and its emphasis on the way in which meaning is generated through the act of performance; and (iv), extending the methods developed in this project to a different repertory, a characterisation of the 'rhetorical' pianistic style of Eugene d'Albert (1864-1932), a pianist whom the music theorist Heinrich Schenker admired, and on the basis of this a re-evaluation of Schenker's highly influential writings on performance. All these materials will feed into a book to be published by Cook under the provisional title In Real Time: Analysing Music as Performance.

An extensive project website intended primarily as a working site for project members, contains among other things a listing of recordings collected for the project, tempo and global dynamic data for many recordings, and analytical visualisations. These materials are freely available and we shall be interested to hear about other uses of them, but ask that you acknowledge the source in any resulting publication.

Click here for a listing of the publications arising from this project.