Acceleration & Deceleration

[snyprrr]

I was just listening to the solo percussion piece 'Psappha' by Xenakis and was marveling at his 'scientific' acceleration in one spot. It sounds to me though, that he is not only using metrical means, but also using an old fashioned 'play faster here'; at least, on this performance (Mortensen/BIS) it's obvious that the player ups the notch in a couple of places during this acceleration (btw- it's the acceleration towards the beginning, before 'long' notes appear).


I'm working on my own bit, going from a quarter note to constant 36th notes, and it's proving quite the task. I've got some parts sounding smooth, but the challenge ahead appears daunting. though... by starting at a quarter note, I don't really have all that far to go. I expect the whole section to last... eesh, hopefully around 16 measures, though I'm calculating for 24-32-36 on the outside.

Is there a proven formula for which note values go in the fourth measure, which note values go in the ninth measure, and so on? Is there no easy way to calculate accelerations,... must I do it note by note, measure by measure, endlessly experimenting? I'm shooting for a fairly smooth transition,... oh, and, I'm limiting myself to 5 chromatic tones, so, the section would sound like five balls in a cylinder that starts moving faster and faster. You know what I'm saying?

Anyhow, what are your thoughts on the technique? Strictly metrical, strictly rubato, or a combination? (as a side note: I've never really found Carter's 'metric modulation' the best way,... huh?,... what do you think about his technique? Examples?)

I do hope this is a topic of interest.

[petrarch]

What you need, if you want it to sound like a real bouncing ball, is to apply an exponential decay. That means the times between bounces need to be decreasing logarithmically. If you want this to be perfectly smooth, these times will be varying continuously, from one bounce to the next, and not simply from one bar to the next. Now, notating this is troublesome because this continuous variation is likely to not fit neatly into a coarse grid like that of a regular sequence of bars. You can either have a bar-less section that lasts however long it needs to, or you write a sequence of bars with changing time signatures in such a way that your bar lines align with the notes--the latter might be useful if you want to have synchronization between instruments. In any case, there is also the challenge that the notation might be too rigorous, making it difficult for instrumentalists to follow precisely. Therefore, some compromise between what the exponential decay formula gives you and something more organic like using accelerando marks is probably enough, for example calculating what the exponential decay would be at the start of each bar, filling it with the corresponding rhythmic value, and then marking it accelerando, with an explicit note stating that the rhythm should be changing smoothly between bars.

[snyprrr]

What you need, if you want it to sound like a real bouncing ball, is to apply an exponential decay. That means the times between bounces need to be decreasing logarithmically. If you want this to be perfectly smooth, these times will be varying continuously, from one bounce to the next, and not simply from one bar to the next. Now, notating this is troublesome because this continuous variation is likely to not fit neatly into a coarse grid like that of a regular sequence of bars. You can either have a bar-less section that lasts however long it needs to, or you write a sequence of bars with changing time signatures in such a way that your bar lines align with the notes--the latter might be useful if you want to have synchronization between instruments. In any case, there is also the challenge that the notation might be too rigorous, making it difficult for instrumentalists to follow precisely. Therefore, some compromise between what the exponential decay formula gives you and something more organic like using accelerando marks is probably enough, for example calculating what the exponential decay would be at the start of each bar, filling it with the corresponding rhythmic value, and then marking it accelerando, with an explicit note stating that the rhythm should be changing smoothly between bars.

You lost me at 'What you need',... no, haha,... no seriously, yes, thank you.

The piece I'm working on is for a solo guitar, and I'm limiting myself to 5 notes, so, those 5 notes are being tumbled around and around, faster and faster.

In some Xenakis he does this process very slowly, and sometimes he seems to 'jump cut' through the process to make for a shorter section.

When I said 'smooth', trust me, I AM allowing for some roughness,... I can tell you that my innate laziness looks at this 'purity' and wants to go back to bed, haha!

I'm developing a 'hiccup' theory whereby I can bypass overt smoothness for a more jarring, yet appropriate, effect. I hope this short section doesn't take me down! Onward, forward!!

[snyprrr]

Seeing as I might be working with a crippled instrument (me!), the task at hand, though still basically the same, is opening up new opportunities for solutions. I am quite excited about this breakthrough.

I've basically been playing the same short section of the line over and over... and over, and... for about three or four days now. The daunting thought of doing a 'pure' acceleration has led me to the 'hiccup' theory in which I have introduced a kind of 'record skip' that has allowed me to do a kind of irrational acceleration,... I'm kind of proud of this creation so far (what?, it's only a few measures), and I see it being firmly at the shorter end of my initial length trajectory.

Ultimately, this experience is reminding me of metal drums, and the double basses, and how many of them use these time tricks to make these crazy rhythms. You know?