|
Click here to go to the Taufactor home page So, Anything musical can be quantised right? I’m a stickler for programming, if someone says “you can’t program feel” that’s like a red rag to a bull with me, even if it takes 3 weeks, it can be done! Or so I thought… During the programming of the Pink Floyd backing tracks for “In The Pink” I found a startling anomaly that appears to be a crack in the Midi spec that I’ve never come across before. There are several sections where this becomes apparent, the most obvious is the syncopated "Triplet" feel Kick Drums in the Meddle track "One of those Days". The basic signature of the song is 4/4 with a shuffle feel. But these runs are Tripleted parts of the shuffle. These sound perfectly natural but in fact no quantise setting works It’s only when I started to think about it, it became apparent that in actual fact the current Midi system is incapable of resolving these correctly. The very basis of the Midi resolution I guess was based around the old Roland Din sync standard of 24ppqm which is the lowest denominator between 12th Triplets (8ppqm) and 16th’s (6ppqm) but for these runs I need 12th (Triplets) "Tripleted" or 8/3ppqm. Which as you can see does not resolve, because all sequencers seem to be based on multiples of 24ppqm (Cubase SX defaults to 480ppqm), the error is less the higher the resolution, but there is still an error. So what to do about 8/3? Or
2.6666666 recurring? The other tricks used in the past, like programming in double time or weird signatures, don’t work either (apart from the fact that these techniques generally make programming the “straight” parts of the song a nightmare). The only way I found of doing it was round the fraction off and manually draw in the note with figures worked out with a calculator. In this case it turned out to be every 293.3333 recurring clicks. This does look like a flaw
in the basic Midi spec to me. Copyright www.taufactor.biz 2002 |