Polyphony Bug, Or Just Me?

[sparkse]

Hi there,

Something has been bothering me about the Wavestate's polyphony modes and while it's super easy for me to reproduce it, I never see anybody else talking about it.

There is, to my mind at least, a bug in the way voices are allocated when you have selected a static number of voices for a program. To reproduce it I follow these steps:


1. Load init performance
2. Go to Program Setup (Perform, Right 4x) and set Hold to ON
3. Go to Voice Assign (Right 1x from last step) and set mode to Poly, Max# of notes to 3
4. Play a 3 note chord.
5. Play another 3 note chord.

Without fail, whilst all the previously sounding voices will cut off as expected, only a single note of the second chord while sound.

This only happens when the voice allocation is maxed out. EG If you have only one voice sounding and play a chord, the sounding voice will cut and then all three voices of the chord will sound. You can test by repeatedly playing chords and hearing how it will alternate between all voices and a single voice.

Unfortunately this severely limits the useability of the static polyphony setting! Since I have never seen anybody else talking about this... Are other people able to duplicate this behaviour?

edit: just to clarify, the bug only happens in the following situation
a) static voice allocation is at maximum
b) the incoming chord is the same number of notes as the max voices

[LosAngeles]

I get the same result. Oddly enough, the voice stealing display doesn’t show any voices being ‘stolen,’ which I feel it should be..

[Bozzy]

Wavestate, and it seems also Modwave (maybe also Opsix?) definitely suffer from voice allocation problems.
It occurs quite frequently that a still being held key's voice(s) will get "stolen" from new played keys, while there are other "free" voices from already released keys still playing their release phase.
This has no sense to me. Released key's voice(s) should be cut-out before other actively playing ones (keys still being held down). This renders playing a drone quite impossible.
It's very strange for Korg, which builds synhesizers since the dawn of time, to make mistakes like that. There must be a reason... but which one?

[sparkse]

edit- doublepost

[sparkse]

Wavestate, and it seems also Modwave (maybe also Opsix?) definitely suffer from voice allocation problems.
It occurs quite frequently that a still being held key's voice(s) will get "stolen" from new played keys, while there are other "free" voices from already released keys still playing their release phase.
This has no sense to me. Released key's voice(s) should be cut-out before other actively playing ones (keys still being held down). This renders playing a drone quite impossible.
It's very strange for Korg, which builds synhesizers since the dawn of time, to make mistakes like that. There must be a reason... but which one?

Interesting that the modwave has a similar problem! Would be great if this could get looked at as it makes it really unreliable to use any polyphony setting than dynamic or mono. I really would like to be able to, for example, restrict my juno-like synth patches to 6 voices. Not just "more authentic" but for instance limiting the number of release tails helps keep a pad sound cleaner. As it is it's practically unusable since you run the risk of accidentally cutting all notes off bar one, when what you intended was to play a chord! Other uses for me include managing polyphony on complex performances that might include chords, leads and rhythmic elements.

Sadly I've mentioned it in a few places now and it never seems to get any traction, which strikes me as weird because it's such an obvious glitch.

[danatkorg]

Wavestate, and it seems also Modwave (maybe also Opsix?) definitely suffer from voice allocation problems.
It occurs quite frequently that a still being held key's voice(s) will get "stolen" from new played keys, while there are other "free" voices from already released keys still playing their release phase.

That shouldn't occur (literally - the software doesn't work that way). If you can provide a repeatable example, I'll check it out!

[sparkse]

Wavestate, and it seems also Modwave (maybe also Opsix?) definitely suffer from voice allocation problems.
It occurs quite frequently that a still being held key's voice(s) will get "stolen" from new played keys, while there are other "free" voices from already released keys still playing their release phase.

That shouldn't occur (literally - the software doesn't work that way). If you can provide a repeatable example, I'll check it out!

sorry to hijack this hijack.. but were you able to recreate the issue described in the original post?

I have no problems easily reproducing it on my end and can provide a video of how to do so if necessary. One thing I will note is that it does not show up on the display as voice stealing, but you can verify that the number of notes sounding is incorrect - eg., performance note allocation is set to 5. I play a chord of 5 notes, 10 voices are sounding. I play the next chord of five notes, and while voice stealing remains always at 0, only 2 voices will sound - a single note.

[BlackForest]

I'd also like to switch into this here - as I am seriously considering getting me a Wavestate..

[danatkorg]

Wavestate, and it seems also Modwave (maybe also Opsix?) definitely suffer from voice allocation problems.
It occurs quite frequently that a still being held key's voice(s) will get "stolen" from new played keys, while there are other "free" voices from already released keys still playing their release phase.

That shouldn't occur (literally - the software doesn't work that way). If you can provide a repeatable example, I'll check it out!

sorry to hijack this hijack.. but were you able to recreate the issue described in the original post?

I have no problems easily reproducing it on my end and can provide a video of how to do so if necessary. One thing I will note is that it does not show up on the display as voice stealing, but you can verify that the number of notes sounding is incorrect - eg., performance note allocation is set to 5. I play a chord of 5 notes, 10 voices are sounding. I play the next chord of five notes, and while voice stealing remains always at 0, only 2 voices will sound - a single note.

By "Performance note allocation," do you mean Program Max # of Notes? Does this happen only if you set Max # of Notes to something other than Dynamic?

[sparkse]

By "Performance note allocation," do you mean Program Max # of Notes? Does this happen only if you set Max # of Notes to something other than Dynamic?

yes and yes, exactly.