Karl A. Petersen <kap@firedragon.com> wrote about (email addresses for)
persons active in WAV->MIDI fields ... good idea! (Karl please put the guy
below on your list).

This is what I found today (no time to check out the web-page):

 >  Paul J Walmsley <pjw42@eng.cam.ac.uk> writes:
 >
 >> In the great WAV->MIDI debate everyone has been focussing on the Fourier
 >> Transform side of things;  as has been mentioned, the Heisenberg
 >> uncertainty principle comes into play a bit with time/frequency
 >> trade-offs.  No-one has really mentioned anything about building models
 >> of musical extracts;  for example, musical signals are known to be
 >> reasonably stationary over a period of around 25ms, which opens the door
 >> to autoregressive modelling, amongst other techniques.
 >
 >> I am researching this very topic for my PhD,  and I would welcome any
 >> comments or suggestions from people who have tried different methods.
 >> Also no-one has mentioned Musical & Harmonic Wavelets which may have a
 >> bit of mileage in them for solving the WAV->MIDI problem;  I can root
 >> out the reference of the paper if anyone's interested.
 >
 >> Paul Walmsley
 >
 >> Signal Processing and Communications Laboratory
 >> Cambridge University Engineering Department
 >> pwalmsley@iee.org
 >> http://www2.eng.cam.ac.uk/~pjw42

Claus Kucher