Analysis of piano performance: Towards a common performance standard?

Werner Goebl
Institute of Musicology, University of Vienna, Austria

The present study concentrates on diversities and commonalities of different expressive piano performances of the same pieces of music. Although each performing musician is able to vary individually and deliberately several parameters of interpretation (tempo, dynamics, articulation, chord asynchronies, pedaling...), one can find many paralleling results.

Performances of Chopin’s Ballade op. 38 (measures 1–45) and Chopin’s Etude op. 10/3 (measures 1–21) by 22 Viennese pianists recorded on a computer-controlled Boesendorfer concert grand piano (SE 290) were analysed in terms of expressive timing, dynamics and chord asynchronies.

Expressive timing is regarded as a multi-leveled phenomenon: at a most detailled level systematic variations of certain rhythmic patterns can be observed (the siciliano-rhythm in case of the Ballade), at higher levels large amounts of slowing down in tempo correspond with phrase boundaries (musical structure). An adequate method of multi-level visualisation is presented.

To make commonalities audible, a numerically calculated average version of all 22 performances was synthesized and played back by the Boesendorfer grand piano. Special attention was given to assure that these average versions sound as natural as the original recordings, so that they become aesthetically comparable without any pianistic or technical artefacts. All parameters of piano expression had to be taken into consideration: firstly expressive timing (note onsets) and dynamics (like Repp 1997) and in addition chord asynchronies were mathematically averaged. Articulation (note offsets) and pedaling were artificially created or added by the author.

These average recordings sound pianistically and musically acceptable and accurate, although they lack individuality and become a bit monotonous in phrasing and tempo design. Averaging over the 22 recordings made sense because of their high similiarity with regard to their timing strategies and dynamic shapings (a PCA delivered only one high loading factor, cf. Repp 1997).

Which aesthetic meaning has this prototype version, could it serve theoretically as a common performance standard?


This paper was presented at the Society for Music Perception and Cognition Conference (SMPC99), Evanston, USA, August 14–17, 1999.


References: Repp, Bruno H. (1997). The aesthetic quality of a quantitatively average music performance: Two preliminary experiments. Music Perception, 14/4 419–444

Acknowledgements: This research is supported by the Austrian Federal Ministry for Science and Transport, in the framework of the START programme (Grant No. Y99-INF).

This page was created in Aug. 1999