Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news4.ner.bbnplanet.net!news3.near.net!paperboy.wellfleet.com!news-feed-1.peachnet.edu!gatech!news.mathworks.com!newshost.marcam.com!zip.eecs.umich.edu!newsxfer.itd.umich.edu!agate!ihnp4.ucsd.edu!equalizer!pfoltz.cray.com!user
From: hpm@bull.cray.com (Hans Mikelson)
Subject: Re: Learning music scales
X-Nntp-Posting-Host: pfoltz.cray.com
Message-ID: <hpm-0707950922530001@pfoltz.cray.com>
Sender: news@cray.com (News Guy)
Organization: CRI
References: <3th1fl$fmp@deimos.rz.uni-osnabrueck.de>
Date: Fri, 7 Jul 1995 14:22:53 GMT
Lines: 25

In article <3th1fl$fmp@deimos.rz.uni-osnabrueck.de>, Michael Saure
<nnmsaure@hal.cl-ki.uni-osnabrueck.de> wrote:

> has anyone made experiences with training a net to classify the scales
> underlying a melody? Are there special resources on the net?

I think a first step would be to get a net that could recognize a major or
minor chord or scale independent of the instrument or composition.  Most
people can do this quite easily but I would expect the mathematics for a
conventional algorithm would be difficult.  I'm not sure how you would
input the sound to the net.

Next move on to other chord types: 7th, sus.  Then have another net which
could recognize the key: C, D, G; given the type of chord and the input
music.  Finally develop a net which can identify all of the notes of a
piece of music by "listening" to it and convert it to Midi format.

Once you have done that you can sell it to Cubase or Cakewalk or somebody
and get rich.

Tschuss,

|    |    |  \   |     /      Hans P. Mikelson
|  __     |  __/ |  \    |    E-Mail: hans@primenet.com
|__  |__  |__    |__ \_  |__  http://www.primenet.com/~hans
