A Repeated Section

Performances of the same musical work do not always play the same amount of music. As an example, we compare two performances of Mozart’s Piano Sonata No. 16 in C major, K. 545, first movement. The performers we compare are Bernd Krueger and Robin Alciatore. Krueger plays both written repeats while Alciatore’s performance leaves both out:

• Krueger: exposition → exposition → development + recapitulation → development + recapitulation
• Alciatore: exposition → development + recapitulation

Audio data: Robin Alciatore, via Musopen (2:14, public domain) and Bernd Krueger (4:23, CC BY-SA 3.0 Germany). Both recordings are used unmodified.

At 0:49.69 on Krueger’s timeline, the exposition starts again. Alciatore’s playhead jumps from the end of the exposition back to the opening. At 3:00.46, Krueger repeats the complete development and recapitulation, producing a second backward jump in Alciatore’s performance.

Repeats in the alignment

Krueger carries the reference timeline because it contains every occurrence. At each repeat sign, two rows share the same Krueger time while Alciatore’s time steps backward:

krueger,alciatore
49.691,51.084
49.691,0.929
...
180.465,133.190
180.465,51.270

The shared reference time makes each repeat explicit in the alignment data. During a repeated passage, Alciatore’s single traversal can accompany either occurrence.

"alignment": {
  "src": "alignment.csv",
  "referenceTimeline": "krueger",
  "timelines": {
    "krueger": "krueger",
    "alciatore": "alciatore"
  },
  "outsideCoverage": "hold"
}

Additional notes

The alignment was computed with chroma-based dynamic time warping. Alciatore’s performance was aligned to one traversal of Krueger’s rendering, then that mapping was unfolded over both repeats.