Soundtrack for NASA LRO rotating moon video
That seductive finger slide

Computer assisted composition, piano preludes

Over the summer, I worked on Moon Ikon, an electronic piece that derives musical gestures based on the contours of the lunar surface.  I'm now starting to work on a set of piano preludes, working title Raw Moon, which are based on the electronic sounds.  I use tools published by IRCAM in the Open Music forum to parse out the electronics, generate  pitch material and export a MIDI file.  From there, I open up the file in Sibelius...where the fun starts.  

If you listen to Moon Ikon, you can hear that the material is very pure, an unrelenting chorus of sine waves.  I've broken the piece in to 28 sections, because I envision 28 preludes and because there are 28 phases to the moon.

ASAnnotation analyzed the sound file, in this case called Moon 19.  The particular view is an analysis of the partials, highlighting those that meet certain threshold defined in the various settings.  I won't make any claims about these values, most of them were default and a few I messed around with just to see if it made any difference ( it didn't).  Here's a visual of the screen.  

ASAnnotation Screen Shot

The output of this analysis is a file in a format called Sound Data Interchange Format.   SDIF is a kind of markup language for music and one of the many cool things it can be used to do is to translate non-musical material into musical vectors (pitch, amplitude etc).  I have a patch in OpenMusic which does nothing except translate SDIF into MIDI.  

A snap of what the  first page of the score looks like when I open in Sibelius.   

Moon Piano 19_0001

If I play that on the internal synth, using a bell sample, sounds like this:

 

Moon Piano 19



 

That's pretty cool, and I have in mind a large scale electronic rendering based on the MIDI files.  Something like The Orb hanging with Jean Michel Jarre on the London Eye scoring up some acid.  But for now, notice that tintinnabulation of the bells is not really represented in the visual score.  I can't account for that, except that the MIDI file is finer granularity than the notation software can represent, but which can still re-play.

 The first edit I did in Sibelius was to expand the single staff onto two staffs.  Then I ran a couple of batch processes to reduce the number of duplicate notes.  I say 'reduce' rather than 'remove'...the SDIF translation generated not just duplicates but  stacks of unison.  That means that when I delete a note because it appears to be a duplicate, I find another identical note underneath it.  When I was done with it, the score looks like this:

Moon 19 with piano_0001

and sounds like this:

 

Moon 19 with piano

 

A horse!  A Horse!  I'd give a horse and a kingdom to know how to tweak Sibelius so that my scores looked like this:

SampleMessiaen

Moving through 28 of the Raw Moon raw material, by the time I finished 6 or so, I was heartily tired of the effort involved.  I couldn't figure out a reasonable spacing of accidentals in a tone cluster, and since I'd already resolved that this was raw material (see 'working title'), I started to ache for a pen and some paper:  in some preludes, yes I would try to recreate this wonderful player piano style, reminds me of the Conlan Nancarrow studies.  In others, the diaphonous trill of the original download.   And in the rest - something like the Messiaen Vingt Regards.  

Comments