Poly-ana: pitchwheel problems?: FIXED!

[dacaumodo]

I'm using the Edirol PCR-M30.

Right now I can't borrow another one, don't know anybody in my area that has this kind of tools.

Having said that, I tend to think the controller is not guilty, because the pitch/modhweel problems occur even when playing back something I recorded, and even with the controller switched off.

Now perhaps using this controller drives Poly-ana "mad" and somehow the synth recalls that the note was recorded with this controller, and hence plays everything wrong after that but I feel I'm treading on black magic territory here

However, if someone using another controller could record a single note with a full-up-then-down-to normal pitchwheel movement with Poly-ana in energyXT 1.4, and share this project with me, I could open it and see if, having been recorded with another controller, it produces the same issues on my machine.

Do you think you could do that MitchK1989?

[MitchK1989]

I'm afraid I don't have energy XT.

I would suggest finding a plain old midi file and importing it to see if it has the same effect as using the controller. Anything that the controller would cause would be recorded when you record the controller stuff.

Could you try drawing notes in by hand and adding pitch bend data as automation with the mouse?

[dacaumodo]

I'm afraid I don't have energy XT.

I would suggest finding a plain old midi file and importing it to see if it has the same effect as using the controller. Anything that the controller would cause would be recorded when you record the controller stuff.

Could you try drawing notes in by hand and adding pitch bend data as automation with the mouse?

I don"t think so, but I'll try. Otherwise I'll ask someone in the eXT forum to record sth for me. Thanks .
d

[dacaumodo]

Update:

more testing was done and described in this thread:

http://www.kvraudio.com/forum/viewtopic ... 38#3048338

SWTrex made some helpful clarificatons and suggestions.

The latest interesting test is the following:


I tried recording a very slow pitchwheel movement (same phrase as before, but much, much more slowly), as you suggested: I get correct pitch behaviour on playback!

We might be on the right track then.

I haven't found a function to thin out CC data stream yet, though. Will dig deeper.

[AdmiralQuality]

I tried recording a very slow pitchwheel movement (same phrase as before, but much, much more slowly), as you suggested: I get correct pitch behaviour on playback!

We might be on the right track then.

Look in Poly-Ana's parameter display field at the top of the GUI. Does it by any chance say "EVENT UNASSIGNED!" (or something like that) when you're having your problem?

[SWTrex]

I haven't found a function to thin out CC data stream yet, though. Will dig deeper.

Sorry...I was thinking of another sequencer when I suggested that eXT1.4 had a midi message thinner. Alas, eXT1.4 doesn't have one.

SWTrex

[dacaumodo]

Look in Poly-Ana's parameter display field at the top of the GUI. Does it by any chance say "EVENT UNASSIGNED!" (or something like that) when you're having your problem?

No it doesn't. When I use only the pitchwheel, nothing appears in the parameter display.
It's only when I use the modwheel that I get "wheel:" followed by a percentage that goes up or down according to modwheel movement. But no error message, ( even though it has happened that the modwheel percentage did not return to zero on release of the wheel but to some random value. This has not happened again in the last two days or so though).

[AdmiralQuality]

Look in Poly-Ana's parameter display field at the top of the GUI. Does it by any chance say "EVENT UNASSIGNED!" (or something like that) when you're having your problem?

No it doesn't. When I use only the pitchwheel, nothing appears in the parameter display.
It's only when I use the modwheel that I get "wheel:" followed by a percentage that goes up or down according to modwheel movement. But no error message, ( even though it has happened that the modwheel percentage did not return to zero on release of the wheel but to some random value. This has not happened again in the last two days or so though).

OK good. Poly-Ana has a very robust MIDI implementation with a very large MIDI buffer. I forget who it was but one of the other developers and I on here were noting that it's possible to overload many softsynths with dense controller streams combined with very long audio buffer sizes. I think whoever it was dubbed it the "64k buffer test" or something like that. Anyway, Poly-Ana has a very large buffer for MIDI events, so it shouldn't be anything to do with the "thickness" of your controller streams. And in the unlikely event there is a a problem, it'll spit that message out in the param display.

I think the modwheel not returning may have just been the behavior of your host and should be able to be worked around by forcing a modwheel=zero message into the start of your project. Also keep in mind that Poly-Ana does something a bit unusual by saving the position of the modwheel in the patch (that's why you see the changing modwheel values in the parameter display) and that patch is (usually) saved by the host-project. So it's possible they can conflict. I run into this as a minor annoyance sometimes when I accidentally hit save when the transport is at a point in the timeline where the modwheel is being performed. When you restore the project it comes back with the modwheel in the position it was when you saved. It's not a bug, was an intentional feature, but it's not always a helpful feature.

The pitch problem however I believe might be a real bug. But it's the kind of bug that's VERY hard to find, as you're the only person who's getting it consistently (or even at all, has ANYBODY else experienced this?) Either that or it's caused by sunspots.

I do believe you though.

End of the month. I'm selling beloved keyboards to make rent, just let the EPS go tonight (but for twice what I bought it for! ). But as soon as I survive this week, I'll be looking at getting a new DAW together so we can start trying some changes ASAP. If this is a bug I'd like it fixed before the 1.12 release, which was about a day or two away from being done when my Dev-DAW died.

BTW, I'm VERY sorry my normal quick response time for support is being affected by my financial situation and the really bad timing on having my DAW go down. If there were more people out there supporting small developers like you have dacaumodo we'd all be better off for it. Thanks again for your patience and understanding while we work through this.

[dacaumodo]

Well I'm sorry to hear about your financial concerns Mike, hope you get rid of them soon.

Thanks for the additional info.

I'm glad if I can help hunting that bug down... and have contributed towards the purchase of your new DAW (or rent!) by buying a license .

I wish there were more people using Poly-ana with eXT 1.4, would be easier to solve (and more "profitable" ) the problem. Although I feel I'm alone in having this problem (I haven't yet read a report from a regular Poly-ana & eXT1.4 user on this forum though, I think), I'm glad you believe me
In case you missed it, here's an example of what I get when playing back or bouncing a note with a pitch bend up one octave, back to initial note, then down one octave, then back to initial note (the pitchwheel range on the preset is +/- 12 semitones, so I should get full octave up and down):

http://vimcortez.free.fr/tracks/1noteon ... ounced.mp3

(the note in question was kindly recorded by DJ Subject, in an attempt to make sure it was not my midi controller acting funny).

cheers!

d

[dacaumodo]

Update:

I tried Poly-ana in eXT 1.4 on my other machine:

Intel P4
3;00GHz 2 gig ram
XP Pro SP2

Gcard: NVIDIA GeForce 6200

M-Audio Delta Audiophile

I don't get the same problem there.

To be absolutely exact, if I loop a clip that contains a note with pitchbend, after a few loop runs the pitch "drifts" very slightly, ie, doesn't seem to quite come back to normal where it should, and this even though the loop points are well away from the start and end of the recorded pitchwheel action.

However stopping the sequencer, and starting again from the beginning seems to give correct pitchbend playback.

I'll test more when I have the time.

d

[AdmiralQuality]

Good to hear it was different on your other machine. But now you seem to have another issue... are you sure the pitchwheel on your controller is centering correctly? That can actually be a problem on older controllers that don't have a wide enough "detent" in the center of the pitch wheel.

Thanks again for the report dacaumodo. I'll be making every effort to get this fixed in for the next update. (And as you're my only tester for this issue, you'll be the first to have the fix! )

Still need to build a new DAW though to get back to work on Poly. Working on that now...

[dacaumodo]

Good to hear it was different on your other machine. But now you seem to have another issue... are you sure the pitchwheel on your controller is centering correctly? That can actually be a problem on older controllers that don't have a wide enough "detent" in the center of the pitch wheel.

I'm pretty sure, yes, I checked it rather extensively. Moreover the drift that occurs is when playing back the loop, and at first the pitch is correct, it only start "drifting" (very slightly) after a while. I'll double check the controller though, just in case. This one's fairly recent, though (Edirol PCR-M30, less than a year old, well taken care of).

Thanks again for the report dacaumodo. I'll be making every effort to get this fixed in for the next update. (And as you're my only tester for this issue, you'll be the first to have the fix! )

Still need to build a new DAW though to get back to work on Poly. Working on that now...

Good luck

d

[AdmiralQuality]

it only start "drifting" (very slightly) after a while.
d

Just checking, is there any chance that's the "analog drift" feature doing it's magic? Set it to zero and doublecheck. When analog drift is set to full, it can cause notes to be off by as much as +/- 50 cents, maybe even a bit more. (I limited the maximum drift to roughly the point where neighbouring notes might overlap in pitch. I figured that was drift enough! )

There's another unique (I think) thing about Poly-Ana's analog drift vs. the competition that I should have pointed out in the "docs". When drift is first turned on from a zero state, the oscillators are still perfectly tuned, and it takes some time for them to drift apart. Setting it to zero again will instantly reset the oscillators to the correct pitch. You can actually play the drift feature this way... turn it up to 10 and when things get too crazy, back off (which simply scales down the range of the existing drift) or zero it -- or rather, turn it to "OFF", which not only scales the drift to zero, but resets the drifts so when you turn it back up, it starts from perfectly in tune again.

In the case of "Drift", each individual oscillator and filter in each voice drifts independently of the others. (This is why it adds a noticeable CPU hit to turn analog drift on.) But there's also the "Detune" knob to consider. Make sure it's zeroed too for your tests. Unlike analog drift, Detune modifies pitch at the keyboard "control voltage" output, each voice having a slightly different pitch offset than it's siblings, so as you turn it up, the voices "spread out" in tune. Detune has a maximum range of +/- 10 cents error, IIRC. Turning it up will cause a "thickening" in a unison patch (with polyphony set above 1). Polyphonically the same thing happens, but you'll hear the detune in chords, vs single notes, and which particular key/note is detuned how much changes with the polyphonic voice assignment. But there's no drift, it stays exactly where you set it. It just controls how well your pitches are calibrated across each of the 12 voices.

[dacaumodo]

OK I'll double check this. It sounds like a likely explanation for the phenomenon I observed when looping a pitchbent note on my "downstairs" machine. Thanks Mike, I'll get back to you.

d

[AdmiralQuality]

Just wanted to let you know that I'm typing this on my new Core2Quad.

Have another day or two of sofware installing before I can get back into Poly but, the DAW deed is done (Ouch!). Sorry about the wait, I expect I'll have something for you to test soon dacaumodo.

Thanks for hanging in there!