Piano Performance Timing
By Robbie Rhodes
Welcome, Mr. Bruno Repp. You said in Digest 95.11.21 that you have performed analyses of expressive timing of piano performances from acoustic records and Midi files. Is there an efficient method available now by which can one can gather the statistics of the near-term performance timing from a phono record? By this I mean when, and in what context, does the artist delay or advance striking the note with respect to the chord and/or the average metronome beat.
Artis Wodehouse has added expression and "broad" timing changes (modest accelerations and decelerations) to the old piano rolls made by Jelly Roll Morton, but they still sound un-natural because the note timing is "too perfect". In the style of the day (1923), the editors removed all the variations in timing in order to make a metronome-perfect "Fox Trot" roll for dancing. In comparison to the rolls, the phonograph recordings by Morton of the same same songs display the normal small variations from the perfect metronome, which we can hear by ear, and which we want to apply statistically to the Midi file of the music roll.
Pushing the "Randomize" button on the Midi editor just applies random phase noise to the performance, and that doesn't sound good either -- these small timing variations are dependent upon the musical context. How should one approach this problem? What can you advise?
-- Robbie Rhodes
|
(Message sent Fri 24 Nov 1995, 05:34:41 GMT, from time zone GMT-0800.) |
|
|
Mechanical
Music Digest™ Archives
