[Resolved in 2.05B4] stuck note! was: aftertouch bug? (or general modulation)

[Xavier]

i will try to reproduce the problem with a puredata patch on linux and osx, do you know puredata (http://msp.ucsd.edu/software.html)? if so i could share the patch and you can try it on your linux machine as well.

Would be great. Let me know if you find something.

[lokki]

hi xavier, find attached a puredata patch that makes the preenfm glitch on linux. the startup values make it glitch after about 10 seconds, enable aftertouch (via the toggle) and it will glitch almost instantly. happens with all the presets i tried (dx7, preenfmbank) but only if the attack is short enough to actually make the note sound. it seems to be indeed very short notes that trigger the problem (5ms between noteon and off) or if you have aftertouch going on the notes can also be longer. experiment with the patch. do you need any "assistance" with setting up puredata (midi out and stuff) or are you familiar with it?

thanks again for looking into it!

[Xavier]

Thanks a lot for the puredata patch.
I can reproduce it... I'll work on that asap.

(In the meantime don't forget that pressing MENU+BACK at the same time, reset the note engine, so this stops the stuck note..).

Xavier

[lokki]

xavier,

pheew, good to know you can reproduce it. i am not going crazy :-)

[Xavier]

This puredata patch is a gem for debuging voice allocation Thanks again 
I attached a beta firmware that should not play anymore the low frequency ghost note.

Explanation : when a voice is playing and when a new midi note is received, the new note does not start right away (which would create an audio "click").
So the preenfm2 use some time (between 0 and 200 samples depending of the current envelope status) to mute progressively the current note before playing the new one.
With a very short new note, the preenfm2 receives noteOff while the previous note fades out and before the new note is acutally started.
This is where the bug was.

The fix i have for the moment should work in most situation: in case it receives noteOff before the note actully starts, the new note is never played.
Which makes weird behaviour with the posted PD  patch (increase a little the note duration and it's fine) but that should work in most real life situation.


Xavier

[lokki]

thanks a lot! will try as soon as possible.

also, just a thought...you could make this fadeout process an option, some people really like clicky envelopes :-) (for some bass sounds it is great)

[lokki]

ok tried it and i can confirm no more stuck low notes, great! but (there is always a but sorry) i get stuck notes on the ones that i play. it happens once in a while (also with the pd patch) it is not such a bad issue, because you can kill it by playing some more notes (but interestingly they have to have a different pitch).

so half way there i would say :-)

thanks!

EDIT: playing with it some more, i get stuck notes also that i cannot stop anymore...have not found a pattern though.

[Xavier]

Yep, it's worse than before 
I'm still on it.

[Xavier]

If you have time to try this 2.05B3....
That was an other problem of too short notes, when the preenfm2 received the noteOff just after the noteOn without starting the audio between.
These too short notes are now also ignored.

Thanks for your help,

Xavier

[Xavier]

Lokki's latest message before split :

"xavier, success!!! i cannot make it glitch anymore :-) will play some more with it, and let you know if there is something unusual, but it seems pretty stable to me."