Unison inconsistencies (bug or natural? Synth v1 RC4)

Bug Reporting
Post Reply
dj.tuBIG/MaliceX
Posts: 61
Joined: 02 Oct 2007 08:18

Unison inconsistencies (bug or natural? Synth v1 RC4)

Post by dj.tuBIG/MaliceX » 10 Oct 2007 12:41

1. KarmaFX Synth v1.00 RC4
2. Cantabile Lite 1.2 (doesn't matter)
3. After some extensive fiddling around with the synth yesterday, it seems I've noticed some 'gradual' issues that happen to the output when the "Unison" module is within the modular workspace.

Problem:
It appears to have an inconsistent phasing that's highly audible, particularly on initial load of a patch that uses not too much detuning. As one plays around over time, the phasing of the unison voices become "consistent" after a minute or so. (The heavy flanging stops.) This may be a problem in cases of bouncing audio, where the unison phase is highly 'condensed' until after a minute of usage.

4. Here's the patch using. though I'd rather you listen to the MP3. The recorded example demonstrates the following:
---Playing a line with the patch already loaded for a while. (no changes done at all.)
---Playing the same line with the patch just reloaded. (notice the difference?)
---Playing at random from that point until the sound becomes normal. (the phase/output tone does not change from that point.)

http://djtbmx.soulbreakers.com/junkpile ... 20Lead.kfx

http://djtbmx.soulbreakers.com/junkpile ... arison.mp3

Question (Seperate issue):
I'm not quite understanding the concept of "Symmetric" and "Straight" settings, however I've noticed that they're the same "pitch" on initial usage. But over a long period of time, either Symmetric would detune down, or Straight detune up. I can't address this as a problem, but I was wondering if you could shed some light on this one. ;)

Thanks in advance.

karmafx
Site Admin
Posts: 1190
Joined: 07 Mar 2005 16:37

Post by karmafx » 11 Oct 2007 22:03

symmetric means that frequencies are smeared symmetricly around the center frequency, i.e., there will both be lower frequencies and higher frequencies in the resulting unison output.

straight, means that the frequencies are only smeared straight up, so only the same and higher frequencies are contained in the resulting output.

I think I was experimenting, and kindof liked both methods.

(I anyone has suggestions for more intuitive naming, just let me know.)

karmafx
Site Admin
Posts: 1190
Joined: 07 Mar 2005 16:37

Post by karmafx » 11 Oct 2007 22:09

I couldn't listen to the mp3 for some reason. :?:

...but the phasing problem you describe sounds like the result you will get if trigger is enabled on the Osc1 (or Osc2) module.

Every time you trigger a key, the phase of the waveform will reset and it does this in all voices (incl. unison voices), so you can get quite large outputs, and it will take a while before the frequency-drift has smeared out the voices sufficiently. You can get around it by disabling "trigger" on the Osc.

dj.tuBIG/MaliceX
Posts: 61
Joined: 02 Oct 2007 08:18

Post by dj.tuBIG/MaliceX » 12 Oct 2007 04:56

I'm pretty sure Trigger's switched off for me, but I'll check when I get back.
Also this only happens when one 'loads' a patch that uses unison.

Thanks for the info on the detuning types, however is it supposed to detune the 'entire' sound upwards/downwards over long periods of time?

As for the MP3, it works for me. (ignore the [url] stuff; it's just an indicator that it's the entire url between.)

karmafx
Site Admin
Posts: 1190
Joined: 07 Mar 2005 16:37

Post by karmafx » 12 Oct 2007 08:08

ok...i will check out your patch later today.

karmafx
Site Admin
Posts: 1190
Joined: 07 Mar 2005 16:37

Post by karmafx » 12 Oct 2007 21:52

Now that I think about it, what you say makes perfect sense when trigger is disabled. When the patch is loaded, all phases are reset to 0, so the patch will sound like it has been triggered the first time you use it, and it may take some time for it to spread out.

I might consider initializing (non-trigger-enabled) OSCs to some pseudo random phase to get around this problem, but I hardly think this is an issue that needs to be fixed for version 1.0.
Let me know if you agree, or not?

I'll check that patch now :-)

dj.tuBIG/MaliceX
Posts: 61
Joined: 02 Oct 2007 08:18

Post by dj.tuBIG/MaliceX » 13 Oct 2007 01:20

Sounds like something worth trying, but yes it's not a 'serious' issue but it is noticeable when designing patches and switching back and forth.

Maybe some kind of extra delay on patch load when calculating the 'optimal phase spread' may help, course this could kill fast patch-switching with GUI closed.

Thanks

karmafx
Site Admin
Posts: 1190
Joined: 07 Mar 2005 16:37

Post by karmafx » 13 Oct 2007 20:13

hey, I tried the patch, and have to agree. It is kind of annoying. :?
So I've now made it so phases are randomized when Osc1 and Osc2 are initialized, and that works nicely. :D
Will be in RC5.

dj.tuBIG/MaliceX
Posts: 61
Joined: 02 Oct 2007 08:18

Post by dj.tuBIG/MaliceX » 14 Oct 2007 01:08

Mad, thanks a bunch.

Really, I'm gonna try to donate for the registered version though I'll have to do it through a friend cause I don't have a Paypal account.

Take care. :)

Post Reply