Hi,

if I understand it correctly Pd for some reasons ignores midi note off messages and assumes instead that any midi note event with a velocity = 0 is a note off.
Now I happen to have to work in a space where they have a midi master keyboard that sends note off velocity and obviously my stuff doesn't work any more as my patches cannot detect a note off, all note events have velocity values associated to the message. I don't know how to work around this.

Does anybody know how to convince Pd to recognize a midi note off with a velocity value? The keyboard in question is an awful chinese brand 'thing' without any user control to switch the transmission off. But that's not the point, I assume many other midi master keyboards send note off velocity as well.

J

strange, i also work with a little maudio master keyboard and i have no issues with midi note off events, [notein] sends a bang when i hit a key and another one when i release it again.

Pd registers ALL note events as note on - even note offs from an external source cause a 'bang' in Pd. what I do in my patches I pack note# and velocity value and filter out zeros - that's how I understand Pd interprets note off.
However if the master keyboard sends release velocity then Pd does not detect zero velocity and hence no more note offs.
Check your midi values, if your master sends off velocity then Pd wouldn't stop ringing... Then how do we tell Pd to shut up?

ah, okay, i guess your problem: do you use [stripnote] in your patches? if not, have a look at it!
it's not to filter the zeros but to filter the note-off bangs... that's where [stripnote] comes in the game.

stripnote filters out zero velocity bangs, not note off bangs unfortunately. Pd doesn't recognize note off it appears, it only recognizes zero velocity note events and assumes that they're note offs. if your master keyboard transmits note off velocity (which is MIDI standard btw) the note offs pass because they're not non-zero.
In my patches I use stripnote to take off the zero velocity note offs as my cheapo plastic keyboard at home doesn't transmit note off velocity (it sets it uniformly to zero), but it doesn't work with a 'better' master keyboard because note its note off is not zero velocity (when I lift my fingers off the keys in slow motion I still get values of ~20).
I should write to Miller... )

You can try [midiin] along with cyclone's [midiparse] to try and get note-offs. You'll likely get an error saying [midiin] only works in Linux, but I think that might not actually be true anymore.

[midirealtimein] says that it only works on windows also, but it certain doesn't!

Sorry, I still am a bit confused by this one, but I think some kind of brute force mechanism might need to be employed here

Maelstrom,
[midiin] and [midiparse] work very well under os x, not even a warning on the canvas. (I refer to Pd extended 0.41.4 and OS X 6.1). Thanks a lot for the hint. Boonier, [midirealtimein] results in an error message on the canvas (though the object appears correctly) and it really doesn't work. For me, at least. Does it work for you under OS X or do you refer to Linux?

Unfortunately also [midiparse] identifies midi note events without distinction of on/off, so the problem to me remains: the only way in Pd to make the difference between note on and off is to identify midi note messages with velocity = zero. And as the master keyboard NEVER sends zero velocity...
If only I could write C, it must be no big issue to write an external that makes the difference between note on and note off messages, it is a different status, I even know which nibble is different.

@ultrasonic said:

Boonier, [midirealtimein] results in an error message on the canvas (though the object appears correctly) and it really doesn't work. For me, at least. Does it work for you under OS X or do you refer to Linux?

no I actually meant OS X, sorry I just reread and realized IO wasn't very clear! It works fine for me - hook up a print to the left output and feed Pd a midiclock (make sure the busses are setup). It will show 248 for ticks, 250 for start and 252 for stop.

It's not a biggy I guess, but throwing errors up in the post window can be misleading.

Theres [xnotein] and [xnoteout] from cyclone if you have Pd-extended

Steffen,

indeed it is the object that should do it but so far it doesn't work, I have no data output from the object at all, neither under OS X nor WinXP. Do you know what data is expected on the inlet? I tried packing atoms from the [notein] object without improvement. The help file messages work, ok, but not 'real' midi input. Also, the values in the first atom help messages are unexpectedly above 127.
Do you know how to use this object? Cyclone doesn't have any further info. In the meantime, I will try to find an answer at the cycling74 site...

Oof I found it - [xnotein] requires raw midi data input, so it receives it from [midiin] at the inlet. I never thought that my Max manual would be this useful one day
Again a small step ahead...
Thanks for the time everybody in this thread!

J