posted 13 October 2012 11:34            

"When learning MIDI continuous controllers, Kyma will now use the more common 7 bit controller message instead of the 14 bit controller messages it was previously using. "

Why is this? Isn't this a step backward?
I understand that more people use 7Bit. But before you switched the learning mode Kyma was able to learn both...

14 bit controllers are downward compatible to 7bit but not visa versa.
It is really said ... since i make heavily use of the more precise 14Bit controllers together with Kyma it feels now like you give a kid some candy and take it then away again ;-)

It would be nice to learn both...or to have a switch in the MIDI preferences to switch the learning mode between 7 and 14Bit.

Let me know if this is possible.

Have a nice day and all the best from Berlin
Chris

posted 13 October 2012 11:55            

quote:

---

Why is this? Isn't this a step backward?
I understand that more people use 7Bit. But before you switched the learning mode Kyma was able to learn both...

---

We changed it in response to a request from someone who uses 7-bit controllers. We are looking into how to detect the number of bits for a future release.

Christian, when you make small changes on your controller, does it send only the least significant bits? Or does it always send all 14 bits? Thanks.

posted 13 October 2012 14:51            

What was the problem with the 14Bit messages?
"Learn" always worked fine with all of my controllers regardless of 7Bit standard CC or 14Bit...

cheers
christian

posted 13 October 2012 15:17            

quote:

---

"Learn" always worked fine with all of my controllers regardless of 7Bit standard CC or 14Bit...

---

posted 14 October 2012 08:48            

Maybe that helps to distinguish if a 7 or 14 Bit message is arriving:

If i observe the midistream from a 14Bit message Mac OS X assigns the "package" of the two control messages the same timestamp for the arrival.

With a 7Bit controller every message has its own timestamp.

Another way would be to check if two consecutive controllers messages types are exactly 32 numbers apart from each other and only then assign
a 14Bit range instead of a 7Bit.

Have a nice day
Chris

posted 18 October 2012 10:09            

As much as the MIDI standard suggests that 14-bit CCs are "official" they are not specified sufficiently to work in the general case. For example as already mentioned there is no rules about whether MSB or LSB messages are sent first. The MIDI standard even implies that you only need send one of the two CC messages for a change. And more directly to the proposed timing detection there is absolutely no guarantee about arrivals. The only real chance you have of correctly interpreting these types of messages requires knowledge of the sender's MIDI implementation.

Another problem is that pairs of CCs used for 14-bit transmission can be interrupted by any other MIDI message. So if the MIDI stream containing the 14-bit values is merged with another stream there is a good chance the 14-bit CCs will arrive with the MSB and LSB separated in time and with intervening commands. This complicates the parser immensely.

Another thing to consider is what happens when the messages flow through another process. For example when MIDI is processed by KymaConnect it has no way of knowing whether a CC is part of a 14-bit pair. So it makes best efforts to deliver the CC to its destination in the shortest possible time. But timing between two consecutive messages cannot be guaranteed (and there is the possibility for a merge). If the MIDI destination is the outbound OSC stream to the Pacarana there is no assurance that consecutive commands end up in the same OSC packet.

So unfortunately I do not think the method will work reliably in general use.