J.D. McEachin jdm at synthcom.com
Wed Jan 29 12:14:35 CET 1997

On Tue, 28 Jan 1997, Arnim X. Sauerbier wrote:

> The bandwidth problems of MIDI were known from day 1.   Continuous 
> controller and sysex data just compounded the problem.  Nowadays everyone 
> wants their sounds to be continuously changing, whereas in the olden-days 
> little more than key-on/off information was required.

But most computer sequencers compensate for this by giving priority to 
note data, which is more time critical.
> To 'override' MIDI would make no sense with commercial instruments - 
> you'd need to design an entire new interface. 

Right - with a digital synth it's completely impractical.

> The standard solution is to parallelize where possible - i.e. get a 
> multi-output MIDI box on your sequencer/computer and run one cord to each 
> instrument/effect - do not daisy-chain them.  Of course, this will not 
> solve the problem of sending too much continuous controller data to one 
> instrument.

No, but with judicious use of data thinnning it eliminates "MIDI Cable 
Clog", and you can still have a lot of CC data.
> Another option would be to just use analog synths, get a bunch of MIDI-CV 
> boxes, each with it's own physical MIDI line to your sequencer/computer.  
>  This way you get very fast, sequencable control over every C/V-able 
> parameter on your analog synth.

But the cost would be enormous.  I believe the most CVs available in a 
single box right now are the Doepfer box.  Say you hook it up to a 
Jupiter6.  You've got note and filter, but nothing else.

> Oh yes, another tip...  Continuous controller data is not truly 
> continuous, rather it's a series of discrete values.  Your sequencer 
> stores many continuous controller values per second, and sending multiple 
> controllers over one channel or line can easily cause overload.  In cases 
> where you're trying to send too much controller data down a MIDI line, 
> try reducing/quantizing the individual continuous controller values in 
> your sequencer, if possible.


On Wed, 29 Jan 1997, Karl Helmer Torvmark wrote:

> In my opinion, the raw speed of MIDI is not the biggest problem.
> On most instruments, the delay from the synth receives the MIDI message 
> to it emits a sound is a bigger problem. This compounds when the synth
> receives a lot of messages at once. The synth manufacturers try to 
> minimize costs, and they choose a processor that can JUST carry the load 
> (of course they try to wry as much polyphony out of it as possibe).

Bingo!  Most analog polysynths are woefully underpowered (and some
digital, like the D50/D550, where the LFOs slow down under heavy MIDI
load!).  Usually it's compounded by bad software design.  We've sped up the
JP6's MIDI response considerably just by writing a more efficient MIDI
message state machine and giving interrupt priority to note data. 
> This is of course a non-issue if you have an analog synth connected with 
> a MIDI to CV-converter with enough processing power.

MIDI is never going to give you the bandwidth that a patchable modular 
synth gives you, but with a modular you're never going to get the 
repeatability, automation, and  control that MIDI gives you.  It's a 
tradeoff.  I'd rather have each and exploit their strengths, not fret 
over their weaknesses.  If you are aware of the weaknesses, though, you 
can usually find a workaround.


More information about the Synth-diy mailing list