Emu XL-7 & MP-7 User's Group

On 12/18/02 7:23 PM, "jesse_medway <medway808@...>"
<medway808@...> wrote:

> Can you elaborate on what makes the quant/swing functions more
> musical?  I hear people talk about this all the time, yet I think
> they are talking more about the timing than the actual math used to
> move notes.
> 
> I used to use an MPC3000 and imported 16th note swing settings into
> cubase and didnt see anything magical going on as far as note
> placement.  
> 
> How is 16th note quant, or 54% swing on an MPC any different than
> other devices/software?
> 
> Jesse

Well I was referring to the net result, how it sounds and in some ways how
it adjusts.  

What I notice most is not how it sounds per say but how the MPC seems to be
easier to drop the groove into a musical-feeling pocket.  In other words
when I dial in swing it "finds" a good sounding value easier.  I'm sure that
sounds like rubbish!  Let me take a stab at explaining why this might be so.

The MPC3K used a 96 PPQN  (parts per quarter note) timing resolution system.
Sixteenth note swing is basically altering the timing of the middle 16th
note between two 8ths.  Smack dab in the middle is 50% swing, 75% swing
delays the note to the next 32nd note position and so on.  Basic stuff.

With a 96 PPQN system each 8th note is 48 ticks.  That means the resolution
for positioning the middle "swung" 16th note is one of the 24 possible ticks
between the perfect 50% position and the "note was so late it became the
next 8th" position.  So the MPC had only 24 theoretically possible swing
values.  In the sweet spot between 50% and 67% there were only 8 possible
settings.

A 120 PPQN system has 8th notes equal to 60 ticks, so there are 30 possible
tick positions.  Between 50% and 67% there are 10 possible places to put the
16th note.  So while 120 PPQN system offers more possible placements for the
note they are not the same as the ones available in the 96 PPQN system.

To illustrate the effect of this say you want to dial in 54% swing.  On the
96ppqn system the closest you can get is 2 ticks delayed, so the swing value
is actually 54.17%.  On the 120PPQN system the closest you can get is again
two ticks away but that ends up being 53.33%.  Clearly not the same.

I picked 96 and 120 for a couple of reasons.  The first is that it
illustrates very clearly the potential differences between two systems.  The
second is that almost every sequencer today works of come multiple of either
96 or 120.

So doesn't a higher resolution sequencer like say Cubase "fix" this issue
because it has more places to put things between notes?  Lets look at what
is happening here.

Cubase, and every other DAW I am aware of, uses 480PPNQ (or some higher
multiple) timing resolution.  That makes an 8th note 240 ticks and gives you
120 ticks of swing placement resolution.  In the sweetspot area (50%-67%)
you have 40 possible placements, 4 times more than 120PPQN (obviously).

How close do we now get to 54%?  Well its 54.167%!  identical to the 96PPQN
result.  The reason this works out so exactly is that 480ppqn is exactly 5
times more resolution that 96PPQN.  So you can perfectly "map" a 96PPQN
recording into a 480PPQN system without any timing errors.

Hmmm...  Seems that would mean that Cubase and other DAWs using 480PPQN
would sound as good swinging as the MPC?  So, assuming that timing
differences are not the whole story what's up with that.

Here's my thinking.  Look at what happens as you dial in swing on a 480 PPQN
system in terms of what you see displayed.  In what follows I am assuming
that the sequencer in question displays swing as a "whole" percentage with
no fractional part.  This is how the sequencers I have used do it (logic,
DP).  I don't know about Sonar or Cubase.

placement  actual swing  displayed swing
126         52.5%            53%
127         52.92%           53%
128         53.33%           53%
129         53.75%           54%
130         54.17%           54%
131         54.58%           55%

What this points out is that on higher resolution systems you can not be
sure with any certainty precisely what timing results when you select a
given percentage swing.  In the case of our desired 54% there are two
possible choices that the sequencer could use.  Which one is picked and what
that choice depends upon is a sequencer implementation detail.

Just for grins this is how it looks for the 384PPQN XX-7:

placement  actual swing  displayed swing
100        52.08%            52%
101        52.60%            53%
102        53.13%            53%
103        53.65%            54%
104        54.17%            54%
105        54.69%            55%

Again two possible choices for 54% swing.  Which is used?

If what I am guessing is true the differences between two sequencer's swing
feel may have to do with the design choices made in terms of resolving the
above issue of which value is used when the user picks 54%!

Even though the MPCs are now much higher resolution than 96PPQN I would
speculate that they have preserved the original swing values exactly.  So
when you dial in 54% your getting 54.166666% as you did on the original
MPC60.

I also think that the values available while you "dial up" a swing setting,
in other words the sequence of possible values you can pass through has some
influence on how the groove drops in the pocket your looking for.  The MPC
swing settings may be more musical not only because of the above effects but
because you can only "find" a limited number of swing settings.  In some
ways it is quantizing the musician's choices to the best sounding ones.

I can not be sure if any of this even has the remotest possible effect on
the legend of MPC swing timing. For all I know it is merely the stuff of
urban (studio?) legend and I too have fallen victim to the hype.

Some final key points.  The fact that importing a sequence from the MPC3K
into Cubase and it looking "normal" needs to be looked at by comparing the
same data "swung" by Cubase using the same percentage settings.  This would
reveal any of the above differences, if they indeed exist. Cubase and any
other high-resolution sequencer should be able to reproduce an MPC sequence
file exactly as the MPC has it recorded.  So the true comparison is not how
the MPC file looks in the higher resolution sequencer but how the same notes
swung with the same percentage compare.

One interesting clue in all of this is that in the groove-quantize presets
in Digital Performer there is a group of ones for the Linn drum machines and
one labeled MPC60/MPC3000.  This gives some credence that there is a
difference.  I dug into the settings for 54% and here is what I found:

This is the sequence of 8 swung 16th notes in a 4/4 measure:
1: 129 53.75%
2: 131 54.58%
3: 129 53.75%
4: 129 53.75%
5: 131 54.58%
6: 131 54.58%
7: 131 54.58%  
8: 131 54.58%

This suggests that the MPC swing timing is not consistent throughout a 4/4
measure.  Is this the source of the illusive swing magic?  And why is it
even like this?  Were these grooves based on actual recorded MPC outputs
(likely)?  Does this mean that the MPC is a little imprecise in its timing
or are we seeing the evidence of Roger Linn's hand?

I can imagine ways that an instrument like the MPC could have been designed
so that the above results were planned and predictable.  Maybe the MPC's
internal timing uses something like that to give the final feel Roger Linn
was after.  Who can really say?

drk

www.delora.com/music
www.mp3.com/zdrk
drk.iuma.com