On an engineering level, what’s most important in determining the timing accuracy of hardware drum machines and sequencers? Can you explain how factors like the resolution of the sequencer and the jitter of the clock affect the timing of the music itself?
Regarding sequencer resolution, the LM-1 – used on all of those early hits by Prince, Michael Jackson, and many others – had a sequencer resolution of 48 parts per quarter note. (48 parts per quarter note permits swing variations of 50, 54, 58, 60, 62, 66, 70 and 75%, and I rarely need more swing increments than this.) The Tempest has a resolution of 96 parts per quarter note but almost all of the great grooves it makes don’t use more than 48 parts per quarter note resolution and often no more than 24.
Regarding clock jitter and specifically MIDI Clock jitter, this can be a factor in software drum machines on older computers, especially Windows computers. But in newer, faster computers and especially Macs running newer drum machine software, it doesn’t seem to be an issue.
Can you tell us a little about how you developed the swing option you introduced on the LM-1?
I discovered both swing and note quantizing by accident. RAM was expensive so in my first pre-release prototypes, I tried compressing the drumbeats by only permitting 16th notes, using one byte per 16th note. When I ran my real-time recording code and played the drum buttons in time to the metronome, I noticed that what I had recorded played back on perfect 16th notes, effectively correcting my timing errors, so I decided to call this bug a feature, which I called ‘timing correct’, which the copycats later called ‘quantize’. In considering how to compress swing-time beats, it occurred to me that this could be done by delaying the playback of alternate 16th notes, and by varying the amount of delay I could vary the degree of swing. And so the swing feature was born, which in 1979 I called ‘shuffle’.
When did you first hear people talking about the magic of the MPC’s timing?
I first heard that idea discussed in the early 80s, shortly after I introduced the LM-1. A part of me would love to tell you it’s something ethereal, a sort of secret sauce that only I can achieve. But honestly, nothing more than what I’ve written in the first answer above is required to create great grooves.
I discovered swing and note quantizing by accident... I decided to call a bug a feature.
And how much of that is actually ‘swing’? A lot of people use ‘swing’, ‘groove’ and ‘timing’ interchangeably despite the fact they have quite distinct meanings.
Swing is the most important factor. You should see the smiles that spontaneously appear when people get a Tempest and simply adjust that little swing knob in real time while the beat plays.
One of the most common cliches about analogue synthesis is that the minor tuning inaccuracies and imprecisions make the sound more appealing. Psychoacoustically, is there any weight to that argument in terms of sequencer timing? Do humans actually enjoy slightly imprecise timing or is it just the more deliberate imprecision of a live musician that’s appealing to us?
Personally, I don’t think anything significant is gained by randomising playing timing. I tried this before my first drum machine and it didn’t help much. And all the great grooves that were created on my drum machines never used any such randomisation.
People talk about the Atari ST having the most precise MIDI timing of any computer-based sequencing platform. Do you think we’ve gone backwards in terms of precise timing as so many of us have moved to DAW-based sequencing?
No. Again, on a newer computer (especially a Mac) running newer software, timing really isn’t an issue.
How does that compare to DAWs and computer-based sequencing approaches? Is dedicated hardware still inherently more precise than a software sequencer running on a Mac or PC?
In newer computers – especially Macs – running well-written sequencer or drum machine software and playing quality MIDI drum pads with good dynamic response, it’s certainly possible to create excellent grooves. But consider that by researching, choosing and configuring these different components, you’re essentially designing your own drum machine. Plus, the likelihood that the dynamic response of the drum pads will match that of the software’s drum sounds isn’t always a given. Further, you don’t get the note repeat (roll) feature with external MIDI drum pads, which I think is very important to creating natural-sounding grooves. Finally, consider that after configuring all the MIDI settings, the inspiration may be gone. I think the computer route is fine if you’re good at the technical part.
On a newer computer (especially a Mac) running newer software, timing really isn’t an issue.
Which way do you think we’re heading in the future? Are people prepared to sacrifice ultra-precise MIDI timing in return for other features or will we see a return of more precise timing?
I don’t see much imprecise timing these days. For the future, I think we’ll see the drum machine continue to evolve as a live performance instrument, as opposed to an offline production editing product. This was my focus with Tempest, the ability to develop highly nuanced personal gestures for real-time drum machine performance. There are some wonderful drum machine players out there with some highly evolved performance gestures.