[sdiy] MIDI HD

Veronica Merryfield veronica at merryfield.ca
Mon Jan 21 08:33:00 CET 2013


Just thought the following might be useful. I am not an MMA member but I have gleaned the following.

HD-MIDI Protocol is the actual name of the new standard.

The working group decided to separate data format ("Protocol") from the transport method. I am not sure what is happening int he transport world. 

Some points:-

* An HD MIDI link is always bidirectional.

* Messages are either 64 bits or 128 bits long (8 or 16 bytes) called "packets". There is a mechanism for variable-length messages. 

* The packets may be sent over any byte-serial medium.

* Essentially any MIDI V1 message function will fit into the short packet including Note-On/Off, per-note controllers, and patch selection all with a lot more channels and resolution.

* A stream of mixed note-on messages will require roughly 3X the data to send. A 100KB/s serial rate would perform as 5-pin MIDI V1 does today.

* Time stamping of packets, which is optional, is included in the packet data format. Thus simply recording the packet data itself is sufficient to capture the entire performance with correct timing to 125uS precision. (Not sure where 125uS really came from, but it seems to point to USB high speed as the favoured transport).

* The short packet note-on has fields for:
  - specifying one of 14 channel groups and one of 254 channels in the group
  - Specifying one of 256 notes
  - 12 bits of velocity
  - 20 bits of articulation, direct pitch, or timestamp
  A long packet has fields for articulation, direct pitch, AND timestamp

* Note Update messages can modify the articulation after note-on

* Controllers are inherently per-note. Note, channel, and device-wide controllers address the "all instances" instance, "all notes" note, and "all channels" channel respectively.

* Controller and patch change messages are atomic (no message pairs or triples needed to perform a single function).

* MIDI V1 messages can be tunnelled within the HD protocol.

Members of the working group are showing devices working with the new protocol at their private meeting at NAMM, Jan 27th.






More information about the Synth-diy mailing list