On Thu, Aug 22, 2013 at 10:37:00PM +1000, Loscha wrote: > MiDI is not ideal for tight timing. > If you look at, say, laying a 4 down kick drum on a track, unless that > tempo you are using is an integer divisor of 31250, it's going to flop > around and not be in time. > Period. That's not quite true. For 31250 to be correct you'd need to be able to time the notes right down to individual bits, and you can't. For a full 3-byte Note On message, you need three bytes - 24 bits, plus six start and stop bits, plus whatever gap between bytes the transmitter inserts - so you've got roughly 1ms per message. Now 1ms is pretty quick but in busy passages where you throw a lot of Note On messages at once you could easily end up with an appreciable gap between the first and last note on the beat. At worst if you fired a Note On message as a full three-byte status on all 16 channels you'd end up with around 15ms of latency between the first and last note fired, which (and you can confirm this with a simple delay effect set for 50/50 wet/dry and no feedback) is enough to cause an appreciable flam. If you bang out a lot of notes on one channel, like a complex drum part, then the problem only gets worse. It's *still* not that bad though, unless you're doing ridiculously busy semi-semi-quaver passages at 160bpm with masses of notes on every channel. If you're up in that territory you want to look at having multiple interfaces. For the rest of us, with our four-part multi and monotimbral JX8Ps with slow-ass envelopes, it won't bother us in the least. -- Gordonjcp MM0YEQ
Message
Re: [CZsynth] Speed of sysex transfer
2013-08-23 by Gordon JC Pearce
Attachments
- No local attachments were found for this message.