[sdiy] Camel*ont* soft Da synth!

Richard Wentk richard at skydancer.com
Thu May 4 20:09:25 CEST 2006


At 18:00 04/05/2006, elmacaco wrote:
>Yes, the Atari Machines are known for both their tight timing and low midi
>latency, I rarely hear about the midi latency because it seems non existent.
>I have a friend with 2 midi peripherals for his STE and he gets something
>like 8 midi outs and 6 ins, maybe more, and each one delivers well to this
>day.
>
>Might be something about having midi functionality being a high priority and
>having it right on the motherboard that makes these things still hang well.

This mystique is nonsense. There's nothing clever about the Atari MIDI 
port. It's a completely bog standard 6850 hanging off an interrupt line 
*which is shared with the keyboard scanner.* It's not some kind of magical 
uber-musical triumph of engineering. It's not even particularly well 
designed. The only advantage it offered over Amigas and PCs and other 
alternatives was that it was built into the box.

I know because I was involved in writing a professional sequencer for the 
Atari, and we had a lot of fun trying to get the timing down to a 
resolution of 480ppqn.

People here seem to be confusing completely different things. The sources 
of innaccuracy in a MIDI system are:

1. Buffering and lookahead processing in the sender. There are various ways 
to do this, and all approaches have strengths and weaknesses.

2. The limitations of MIDI bandwidth, which creates unavoidable errors in 
any system with:
         i. More than one polyphonic synth on a single MIDI cable chain
         ii. Very, very heavy polyphony for a single synth line - e.g. 
unplayable 30-note chords.
         iii. Heavy use of controller data.
         iv. Poor prioritising/thinning algorithms used by senders - e.g. 
not using running status, not prioritising note events over controllers, 
and so on.

3. Internal latency within a MIDI receiver. Every synthesizer takes a 
different time to parse MIDI messages and trigger the sound hardware. Some 
synths are much better at this than others, and all respond with a 
combination of both fixed and variable delays.

There's also the famous hypothetical MIDI Thru delay, which so far as I can 
tell is only ever a problem for systems that use a software Thru mode 
instead of using a direct hardware link.

And that's it - all there is to know about MIDI. It's actually a fairly 
crappy system that needs some thought to make it work well. If there's any 
kind of latency, that's a design error, and shows that someone has been a 
bit casual about how they've implemented it.

Richard



More information about the Synth-diy mailing list