A measurement of piano playing
How often a player brings out a line that the other players of the same piece leave inside the texture.
The same passage — Nocturne Op. 9/2, bars 10–11 — from the recordings that sit lowest and highest on this measurement. Every clip is a note list replayed on one sampled grand at one gain, so the only difference you are hearing is the playing.
Every note is expressed relative to that player's own melody level nearby; the median across all players is the consensus for that note; a run of notes played well above consensus — after removing the player's own overall offset — is one event. Counted per ten bars. This is the only measure here that is defined against what other pianists did.
Every piece in the library, with the range across its recordings. The units mean the same thing in every row, except for basic tempo, which is in each piece’s own beat.
| Piece | Recordings | Lowest | Median | Highest |
|---|---|---|---|---|
| Bach Prelude in C | 27 | 0.00 | 0.00 | 0.00 |
| Chopin Étude 10/1 | 31 | 0.00 | 0.25 | 0.63 |
| Gymnopédie 1 | 23 | 0.00 | 0.51 | 0.90 |
| Mozart K. 545 | 37 | 0.48 | 1.11 | 2.26 |
| Fantaisie-Impromptu | 20 | 0.80 | 1.18 | 1.81 |
| Brahms Op. 118/2 | 28 | 0.32 | 1.29 | 2.02 |
| Ravel Ondine | 61 | 0.95 | 1.47 | 2.11 |
| Rachmaninoff Prelude | 29 | 0.82 | 1.48 | 2.46 |
| Träumerei | 31 | 0.61 | 1.52 | 2.73 |
| Clair de lune | 27 | 0.83 | 1.55 | 2.39 |
| Nocturne Op. 48/1 | 27 | 0.91 | 1.56 | 2.21 |
| Pathétique II | 47 | 0.96 | 1.64 | 2.60 |
| Chopin Ballade 4 | 12 | 1.34 | 1.67 | 2.01 |
| Schubert Op. 142/3 | 9 | 1.25 | 1.94 | 2.55 |
| Ravel Pavane | 33 | 1.53 | 2.08 | 2.64 |
| Schubert Op. 90/3 | 43 | 1.40 | 2.09 | 3.14 |
| Nocturne Op. 9/2 | 32 | 1.32 | 2.11 | 3.42 |
| Chopin Étude 10/12 | 19 | 0.95 | 2.14 | 2.86 |
A Disklavier capture — the pianist’s own key sensors, no microphone — was rendered to audio, transcribed as if it were a recording, and re-measured. The error the transcription step introduces, as a multiple of how far pianists actually differ on this dial, is 0.51. fair Usable, but the error is worth stating alongside any single value.