Objective performance analysis as a tool for the musical detective.

sures in EllY Ney's recording. The two renditions of measures 5-8, however, were virtually identical, within the limits of measurement error. The evidence is presented in Figure 1. Figure 1a shows the differences between the first and second renditions in the inter-onset intervals (lOIs) between sllccessive (clusters of) tones, as measured in the digitized acoustic waveform (see Repp, 1992). The expression and individual character of a musical performance resides in its microstructure, which includes variation in the exact timing and intensities of the tones played, Each performance has a unique microstructure which cannot be reproduced exactly by a human performer, even though repeated performances by the same artist are often highly similar. Objective performance analysis (Seashore, 1936) thus makes it possible to distinguish an identical copy from a repeat performance with a high degree of confidence. A demonstration of this capability was provided by an unexpected discovery during the analysis of 28 different recorded performances of Robert Schumann's famous piano piece, "Traumerei" (Repp, 1992), The measurements revealed beyond any reasonable doubt that one recording of older vintage, by the German pianist Elly Ney (Electrola WDLP 561), contained two identical sections four measures long. The first eight measures of "Traumerei" are followed by a repeat sign in the score, which is obeyed in nearly all performances. When the expressive microstructures of the first and second renditions are measured and compared, they are usually found to be quite similar in any given performance, but never identical. Since artists are not machines, there is always a certain amount of uncontrolled variation in timing and intensity, in addition to any variation introduced deliberately by the artist. Such variation was also observed between the two renditions of the first four mea40