Today I’m working on some pattern sequencer stuff (the sequencers that sequence the smaller sequences – wow, that is not any clearer). We have three primary synths in the game; Lead, Bass and Pad. The Lead and Bass synths are monophonic (1 voice/note played at a time) and the Pad is polyphonic (currently set around 4-5 voice/notes at a time). The pattern sequencers I’m toying with today are the for the Pad synths, thus they’ll be juggling a bunch of voices.
What’s tricky about the pads (and kind of cool, it was Henk’s idea) is that the Pad sequencers will progressively steal voices as the patterns ask for new notes. This behaviour is a bit like the classic Yamaha CS poly synths (CS-50, CS-80 etc), and with a bit of modulation trickery, we could treat each voice like an individual (and slightly different) oscillator. Long story short, it should SOUND REEL GUD. I’ve posted this sample before, but it illustrates this nicely:
Our current voice budget is 10 realtime voices spread accross the various synths – which in software like Live, Reason, Cubase etc is peanuts (especially when you consider that our synths are agruably ‘lower’ quality). But they aren’t rendering a 3D world with things like physics, laser-beams and fun-emulators. With 6-7 musical voices being used up, it would be fair to ask what the rest are for, and while I’ll save the nitty-gritty for another post, we do a lot of the sound effects (especially the reactive/interactive ones) with synth voices.