Saturday, February 09, 2013

Five-Voice Ricercare for Orchestra

This is another one of those fugue subjects I composed several years back that I continue to make progress with. The subject itself dates from about 2003, and it was composed as a five-part canon at the octave. In 2006 it became a five-voice perpetual canon for string choir, which I blogged about at the time, but since you're going to hear it again, I didn't scour the archives for a link to those posts. Then, just last year, I came up with a two-voice octave fugue on that subject for solo guitar (An octave fugue is what Bach called a two-part invention). That is now the prelude for my Sonata Zero for solo guitar.

When I came up with that solo guitar version, I knew I'd made the breakthrough that would finally allow me to compose the five-voice exposition, and after some fits and starts, I finally got it about 95% there, I think. So, today's version just has the five-voice exposition, the first episodes and interlude, and then the five-voice perpetual canon, which is the recapitulation and codetta. I have a general plan for the development area - go to the winds for duets and then trios for the stretti, and gradually work back up to tutti - but I'm not ready to tackle that yet.

Here is the MIDI to MP3 conversion: Five-Voice Fugue for Orchestra

There are some default chorus and reverb settings in my new MIDI2MP3 converter I'm going to have to remember to defeat in the future, but I really love this application! Far superior to my old iTunes workaround, and since they are MP3 files and not M4A, the Play Tagger app I have imbedded in the blog template should work again too. A win/win.

On the top system is the subject, which is reminiscent of the Royal Theme - probably written by C. P. E. Bach to test his father - that J. S. Bach used for the Musical Offering, but with a different head and tail, so that it makes the five-part canon.

Then, on the lower system, is the answer, which has to be real and not tonal, and the counter-answer, which is the head of the subject in augmentation after measure 7.

The aforementioned counter-answer makes the leap of the diminished seventh at the dovetail into the third statement, which is the subject again. What was the minor sixth degree from the perspective of the dominant region therefore becomes the minor third of the tonic. Very slick and elegant. The countersubject, now in the lead, then gets the augmented version of the head figure.

Here, a conundrum occurs, as it is not possible to use the leap of the diminished seventh directly into a new statement of the real answer. This is because the minor sixth degree coming out of the tonic - the f-natural in measure sixteen - would be the minor second degree of the dominant. Utterly unworkable. So, I had to introduce a four measure episode in the middle of the exposition to reorient the voices for another statement of the real answer. Far from being any kind of fault, this increases the interest of the exposition enormously, avoiding the predictability and monotony that five statements in an uninterrupted row would produce.

The fourth thematic statement - the second real answer - then appears in the highest octave of the orchestra, and the primary counter-answer, with the augmented head figure again, provides the bass voice. This sounds quite dramatic and majestic. The third counter-answer - in violin two - adds a chromatic suspension/resolution sequence versus the subject that is quite beautiful, in a vaguely unsettling way.

Finally, the fifth thematic statement - and the third of the subject - appears in the contrabass octave, and it is positively massive sounding. This was the point of this fugue: To compose a fugue that used all five octaves of the orchestra. I doubt I'll ever compose another fugue of this type, and I also doubt I'll ever go beyond five voices to six or more. While possible for a subject of a more limited compass, five-voices seems like the pluperfect arrangement, as it is also reflected in the five voices of pure harmony (Four voice transitional stratum over a constant-root bass).

I didn't put the countersubject in the lead, as I, 1] didn't want to repeat the melodic peak, and, 2] wanted to get out of that highest octave fairly quickly, since it can become grating.

Coming out of the subject again gives opportunity for a larger version of the previous episode, and that's where I'm able to let violin one bow out gracefully: It sounds like it is absorbed into the E-natural in the second violin part. I've gotten much better with this aspect of fugue writing over the years.

On the bottom system we get the first six-measure interlude - as opposed to the four-measure episode - and it's very majestic and impressive sounding, what with the cellos and basses playing deep, dark counterpoint with each other, and the violas and second violins providing contextual color for them.

Now, at this point in the final version, the tutti will break and the winds will start with the stretti - initially in two voices - but hear I just link directly to the recapitulation, which, as I said, is a five-voice perpetual canon.

Here we get five inexorably descending statements of the subject in octaves and single measures of delay, which is how I composed the subject those ten years back. That means that measure forty-five has all five measures stacked up from the bottom. Pretty gnarly. Starting in the second system those dovetail into the subject in augmentation, which I did not intend: I discovered it back in 2006. Much of the art of fugue writing is, as I tell my students, "Noticing shit." Just becoming aware of relationships that you, as a composer, may not have strictly intended, but which add the dimension of art and magic to a composition.

On the top system in the top voice, you can see that I had to modify the augmented version of the subject: I was able to keep the rhythm, but I had to ditch the chromaticism. I also then dropped the final measure of the subject, and at that point everything dovetails back - re-dovetails? - into the original version of the subject, thus proving the perpetual nature of the five-voice canon. The augmented segment sounds wickedly dissonant, by the way, because I use all of the inversions of the bVI(M7) chord that have the root above the seventh, which yields the interval of a minor ninth. This really only works in five voices - possibly four - but it sounds quite devilish and creepy, which of course I love. lol.

Finally, the piece repeats the dissonant augmented statements, but this time over an ostinato on the tail figure, as the piece comes to an end: As each part reaches the tonic, it sustains that note until the conclusion. Since the subject has eleven of the twelve pitch classes in it - yes, there is an a-sharp in the answer, but no minor second degree is ever heard in either region - I put the final lick as a chromatically descending triplet that presents the b-flat as the penultimate note. Did you really think this weird and haunting piece was going to end on a major chord?

Sunday, February 03, 2013

Apple has Broken iTunes for Me

I guess it was always a workaround, but I had a nice workflow with it anyway:

1] Compose in Encore

2] Export MIDI file

3] Drag/drop into iTunes

4] Select soundfont bank in QuickTime Preferences

5] Create AAC version in iTunes with selected soundfont

Well, starting in OS X 10.6, there are no longer any QuickTime Preferences, and after 10.7 iTunes 11 won't play MIDI files at all. So, I've had to compose on my Mac Pro (10.7.5/iTunes 11), send the Encore file to my Macbook Pro (10.4.11/iTunes 9) via the HD attached to my AirPort Base Station, and do all of the fiddling there. Yes, a major PITA.

I finally had enough of that noise, so I was looking for a MIDI to MP3 converter. Well, those are a dime a dozen, but none of them allowed for selection of soundfonts, as far as I could tell. Finally though, I found a nice app for the chore that allows me to use my soundfonts: MIDI2MP3.

It's completely intuitive. Less than sixty seconds after opening it, I had a test conversion done.

So, if you need to make MIDI files into MP3's using your own soundfont collection, this is the easiest way. It's a major improvement from even my pre 10.6 workflow: Many less steps.