MIDI HD

[Koekepan]

It just struck me:

Don't look for the new MIDI spec to come out soon, and don't look for it to be well-supported, or particularly useful, when (if) it does.

Not because there aren't smart people involved. There clearly are.

Not because they don't want it out. I'm sure that the companies would love to sell us new studios.

Because they're too entrenched, and entangled, to come out with anything elegant, clean and functional any time soon. Every discussion turns into a six months long wrangle about how some obscure synth by a company on life support did things twenty years ago, and how that affects backward compatibility or the desperate need to support some half-forgotten hack that technically was in spec at the time that it was done.

We'd be better off with a brand new, clean spec that incorporates a straightforward clock signal for syncing with old gear running on old MIDI. That's your compatibility layer: the clock.

Given the current lamentable state of constipation, our best chance is likely to be a small company that looks at MIDI, declares it unfit for purpose, and comes up with something clean.

When you come down to it, how much do you really need?

"Timestamp/start/stop"

"Start a note-thing on channel at time XX.XX" -- "Reference cookie YYYY for your note"

"Set parameter ZZZ to TTT on reference YYYY"

"Alter parameter ZZZ by TTT on reference YYYY"

... and so on and so forth. As protocols go, it's pretty lightweight. Compare and contrast with email, for example.

Put it in ethernet frames on 1Gbps CAT6 cables, and your basic hardware is available for cheap, universally, and you can implement units around a Raspberry or an Arduino. These days even Arduino can run a substantial synth if you have a chip with decent floating-point on it. Unmanaged ethernet switches in this speed range are also pretty cheap per port at this speed range, and you could transmit ludicrous quantities of data and audio with very low latency in a single star topology setup (although snowflake topologies would still be viable if you're trying to record an entire symphony orchestra or operatic performance).

In this day and age, it's almost accessible as a hobby project.

In fact, if I were to prototype it myself, I might just stuff the messages into SNMP.