Hi Herbert,
It seems to me you really have two goals: ear training by abstract
interval and style training by context. As for the first, I think you should
make a wide variety of scales available including things far outside the early
music realm. Learning things like octatonic or chromatic melodies is going to
increase one's ability to distinguish melodic sets even though they may not be
directly applicable to the music the person plays. Let the user select which
scales they'll use for the session. An important added feature would be to
include temperament systems.
The second goal is much more difficult to program. I'll just focus on
baroque music because that is mainly what I do nowadays, but the same issues
would apply to a renaissance program as well. Early musicians learned to
compose very much like jazz musicians learn to improvise. (Of course, early
musicians learned to improvise, too.) Today, most jazzers learn by copying
licks from the recorded catalog. A saxophonist might learn a few second of
Miles's solo at 5:13 on "Love for Sale" from Cannonball's Somethin' Else. They
practice it in all keys and can then potentially insert it with variations
whenever the same chord progression pops up in any tune. Then they move on to a
lick by Monk or Trane and do the same. This is not cheating: it's learning
style. In a similar way, baroque musicians worked with a large but finite
library of motifs. This is how Bach could write a cantata every Sunday and
Handel could compose a seven hour in 6 minutes and 52
seconds. It's also why Bach reworked material by Vivaldi, Weiss and others: so
he could learn their licks.
So... if you really want the program to teach users how to hear baroque
music, a random melody generator won't work. You would need to be very familiar
with the structural principles of compositional practices of the era. Then you
would have to code the program so that it knows how to present idiomatic
gestures over, say, a ii-V7-I progression or a sequence by fifths. The
difficult part would have to do it with enough variety to be stylistically
plausible yet not so predictable that users end up learning your program's
idiosyncratic style. The program should be able to distinguish between Italian,
French and German styles as well as the rhythmic stereotypes of the major dance
species. Then there is the whole subject of species counterpoint and polyphony.
I imagine all of this would take some time to build. The same thing would
happen with renaissance music, but with different parameters.
That may have veered wildly out of the range of what you have in mind. I
hope it's helpful.
Chris
Dr. Christopher Wilke D.M.A.
Lutenist, Guitarist and Composer
www.christopherwilke.com
--------------------------------------------
On Sun, 11/24/13, Herbert Ward <[email protected]> wrote:
Subject: [LUTE] Scales for ear training.
To: [email protected]
Date: Sunday, November 24, 2013, 9:03 PM
I contemplate writing an ear-training computer
program. It would select random pitches from
a scale and pair them with random note values
to create a melody, which the trainee would
listen to and try to reproduce (by voice or
instrument). The trainee could ask the program
to replay the melody as needed.
I have no expectation that the melodies
will usually musical. But, leaving that
question aside, my question is this:
For a typical lutenist involved in the Renaissance
and Barqoue repertoire, what scales would be
most useful? Some possibilities would be:
major
natural minor
harmonic minor
melodic minor
dorian
mixolydian
....
Also, if you're interested in using such a program,
let me know what type of computer you use (Windows,
Macintosh, or Linux). If more than a few people
respond,
I'll make certain accomodations during the programming.
To get on or off this list see list information at
http://www.cs.dartmouth.edu/~wbc/lute-admin/index.html
