[LAU] Replacing/enhancing MIDI, and OSC

Hanspeter Portner dev at open-music-kontrollers.ch
Mon Feb 16 22:54:13 UTC 2015

Hash: SHA256

On 16.02.2015 18:36, Harry van Haaren wrote:
> No, the MIDI "cable" baud rate is 31250, while MIDI over USB runs
> at 112500. Its faster - but there are still limitations.

Where do these 112500 Baud come from?
Just curious...

Does this refer to low-speed USB-MIDI (e.g. 1.5Mbit/s)?

112500 Baud = 1.5Mbit/s / 8bit * 3/4 * 0.8

3/4: USB-MIDI prepends 1 byte to every raw MIDI 3-bytes [1]
0.8: -20% USB bus overhead


> Regarding the above: -Sequentially handled: yes

Not necessarily.
USB is packet based and USB-MIDI uses bulk-transfers with a maximal
packet size of 64Bytes. A common 3-byte MIDI instruction packed in a
4-byte USB-MIDI instruction thus gives you 16=64/4 possible concurrent

Whether those concurrent events will be treated as such by a host is
another matter...

high-speed USB has a maximal packet size for bulk transfers of 512
bytes, I'm not sure though whether this applies to USB-MIDI, too?

> -Bandwidth: lots better than DIN-cable MIDI, but limited

Depends on underlying USB speed, I guess
low-speed (1.5Mbit/s) vs full-speed (12Mbit/s) vs high-speed (480Mbit/s)

>> 2.  I had not heard of MIDI-over-OSC.

You can also do OSC-over-MIDI via sysex messages ;-)

> Yes - almost everything is possible with OSC - any form of data can
> be passed around as a "blob" of binary data, timestamps are
> available to be used for timing, and "bundles" can be used to have
> multiple events be processed at exactly the same time.
> Sequential handling? Yes - unless in a bundle. Bandwidth:
> unlimited.

Just like MIDI, OSC's maximal bandwidth is limited by its underlying
communication layer [2]. You may send OSC via UDP/TCP on Gigabit LAN.
But people do pack it into USB and Serial, too.


[1] http://www.usb.org/developers/docs/devclass_docs/midi10.pdf
[2] http://nime2008.casapaganini.org/documents/Proceedings/Papers/163.pdf

More information about the Linux-audio-user mailing list