The more the merrier :-)

I like the idea. The quad-tree decomposition I used has the drawback that pick-up 
notes could render
it useless. Polynomial approximation minimizes that effect.
 
>         3)  Measure the (signed) distance of each successive note
> above/below middle C. Form a vector of these, (v(1),v(2),...,v(n)).
> Think of this as a function for the moment: v(t), t=1,2,...,n.  In order
> to make the result independent of the key, subtract the (weighted) mean
> value from each v(i), and call the resulting vector V.

This is not clear to me. I might be wrong, but by using intervals instead of notes, 
you are making
it already key-independent. So I don't see the use of subtracting a mean. Anyway, even 
if you have
to, I don't see how subtracting a mean would render it key-independent. 

>         (f) Obvious problems: one has to be able to compare tunes in
> different time signatures.  This might come down to re-scaling the
> polynomial. 

I don't think it is necessary to handle differences in time signature.

> Finally:  I suppose anyone suggesting an algorithm should have to
> implement it---fair is fair---but while I could probably write the routine
> from step two on (because it's pretty easy), the task of writing a parser
> to get to step one is a bit much for me. Sorry.

I had to write such a parser to implement the quad-tree decomposition. It's not 
entirely fit to
reuse for something else, but I planned to make it more application-independent 
anyway, so this
might be a good occasion. Would an object-oriented Perl module be of any use to you ?

Brgds,

Mark

-- 
------------------------------------------------------------------------------ 
  Mark Vandenbroeck                Email   [EMAIL PROTECTED] 
  Customer Support                 Phone   +32-2-719.12.11   
  Oracle Belgium                           +32-2-719.59.51 (Support) 
  Vuurberg 80                      Fax     +32-2-725.48.74 
  B-1831  Diegem                           +32-2-719.13.14 (Support) 
  Belgium                          Web     http://www.be.oracle.com/~mvandenb 
------------------------------------------------------------------------------
begin:vcard 
n:Vandenbroeck;Mark 
tel;fax:+32-2-719.13.14
tel;work:+32-2-719.12.12
x-mozilla-html:TRUE
url:http://www.be.oracle.com/~mvandenb
org:Oracle Belgium;Customer Support Services
adr:;;Vuurberg 80;Diegem;;B-1831;Belgium
version:2.1
email;internet:[EMAIL PROTECTED]
title:Principal Technical Analyst
x-mozilla-cpt:;0
fn:Mark Vandenbroeck
end:vcard

Reply via email to