Polyphony and Acoustic Pianos (bug and workaround)

[xp30]

I talked before about polyphony on the Roland Fantom. As mentioned before, if you only know one thing about avoiding audible polyphony issues, then it is to set voice assign mode to SINGLE (see https://forums.rolandclan.com/viewtopic ... 66&t=69782). I have also talked about optimizing preset tones, and that the least suspicious tones might consume the most voices.

While experimenting with these issues, I noticed one more thing: It seems that acoustic piano tones drop out earlier than other tones.

This makes the situation worse, because when layering, the acoustic piano tones are often the dominating parts. It is also surprising, because acoustic piano tones are also just Zen-Core tones (including SuperNatural acoustic pianos).

So, I did a few more tests, and I actually found one feature that is typically used for acoustic piano tones, and that seems to make the difference: Redamper (including half-pedaling).

If I disable the redamper feature on acoustic piano tones, then they no longer drop out first. And vice versa, I can reproduce the issue with tones created from scratch that only differ in the redamper feature. To me this sounds like a software bug, where the importance of a voice with redamper feature is incorrectly calculated/tracked, and then eliminated first.

The good news is: It is easy to disable the redamper feature on Zen-Core tones (and slightly more complicated on SuperNatural acoustic pianos), and it is not an important feature for most non-classical keyboard players.

I will share audio examples later.

[xp30]

I created an example. The following scene uses 4 zones. It uses the same Zen-Core tone in all 4 zones, and this Zen-Core tone is a simple sine wave with redamper setting enabled. The 4 zones contain identical clips (3 measures). The clip uses the damper pedal, however, without re-dampening or half-pedaling. That means it should make no difference whether re-dampending is enabled in the tone or not.

something.zip

Fantom Scene Something

(3.29 KiB) Downloaded 45 times

For the experiment, I disable the redamper setting in zone 2 and 4 (by switching the envelope mode). I then start all 4 clips, however, listen to and record the zones individually.

• Zone 1 (with redamper):

  something-sine-with-redamper.mp3

  (217.31 KiB) Downloaded 60 times

• Zone 2 (without redamper):

  something-sine-without-redamper.mp3

  (218.37 KiB) Downloaded 47 times

Zones 3 and 4 sound identical to zones 1 and 2.

This experiment demonstrates that the zones/tones with redamper setting drop notes earlier than the zones/tones without redamper effect. I intentionally created a very simple tone from scratch to make it easier to understand that it is related to the redamper setting. It does not sound nice. I might post later an audio example with an acoustic piano tone, with and without redamper setting.

[xp30]

Here is an example with a piano tone (SN-AP:EXSN03:0001:Classic Piano), using the same scene/clips as in the previous example. Again, all 4 zones/clips are playing simultaneously, however, only listening to and recording one zone.

• Zone 1 (with redamper):

  something-piano-with-redamper.mp3

  (240.94 KiB) Downloaded 74 times

• Zone 2 (without redamper):

  something-piano-without-redamper.mp3

  (243.33 KiB) Downloaded 78 times

[Old No7]

Wow, very interesting and enlightening results -- and GREAT work and thought were put into this!

That descending piano part "with redamper" sounded almost "toylike" at 8 seconds -- what a difference.

I don't have a Fantom-O "yet", but I'll make note of this!

Old No7

[xp30]

I updated today my Fantom-7 from v3.01 to v3.02. This update has fixed the problem for the big Fantom, and I guess v1.04 is the same for the Fantom-0.

[thortschman]

This is genius. Definitely the bug so many people where complaining. And Roland said there is no such problem.. thanks a lot - I’m also sure latest update will fix it.