Wednesday, September 30, 2020

Prometheus Programming, Timing in Music

Yesterday was a slow day of feeling somewhat ill, the strange burning pain in my throat, chest, ears/nose, which slowly gyres around over a few days, never quite leading to sniffles or coughs, yet always feeling raw and sore. This comes and goes regularly and has for at least 10 years, and when it comes it knocks my energy levels. Just at the point when I feel a cold may fully emerge, it fades away and I wonder if there was anything ever there at all. As a result, I did little yesterday but rest and tweak The Jabberwocky, which is just about complete now.

Today I'm updating my music software, Prometheus. The biggest change is to the splitting of big samples. The program only supports samples of 90 seconds, so for longer samples it can auto-split these into 90-second (ish) sections and calculate, down to the exact sample, the place to start to play the next instrument to continue playing seamlessly.

This works fine, but has never worked when the tempo is toyed with because this really complicates the maths; essentially the program has to start at the start of the song and run through it all so that it can split the sample exactly correctly. I've made this change today. One downside it is that it is a lot slower to calculate than the old simple split. It can take a minute or more to split a long sample up, all due the need to go back to the start of the song and run through every change of the tempo. It can take as little as 5 seconds, or as long as 3 or 4 minutes if placed right at the end of a 200-minute long song (as IF I'd ever make a song that long! - but it is important to test the function in every extreme).

A second change is related. A similar calculation needs to take place when starting to play a sample in the middle. Normally in a sequencer, you click Play and the subsequent notes will start, but any that started earlier are not played. To be 100% accurate, the tail-ends of any prior notes should ideally play, and this can be very important, particularly when those long samples are there or when you start a long drone somewhere back at the start of a song. This is also a complex and slow calculation. The ability to do this accurately at any speed was hidden in the program options, so I've made this more overt and easier to toggle on and off.

These changes indicate that I'm using lots of long samples and am playing with the tempo a lot - in fact nowadays I rarely stick to one tempo throughout a song and try to speed up or slow down things for different moods. Regular tempo is one of the three big mood-killers in digital music, regular volume and regular pitch being the others. Early electronic music (pre-1985) sounds better than later because early synths were not pitch reliable and because digital sequencers were rare and complex to program.

One other thing I do now is craft each note rather than ever use shortcuts or copies. Even drumloops are different each time and not as perfectly accurate at they should be. Timing is crucial to emotion in music because music, like any drama, is all about anticipation and surprise - both are enabled principally by toying with time.