Filter bug in split mode

[davydark]

Hi folks. When I use Polybrute in split mode, there's a bug with the filter.

If I play three notes on side A, then six notes on side B, the first three notes on B sound as they should, but the second three notes sound very different, then the rest sound normal.

If I play four notes on A, then six notes on side B, the first two notes on B sound normal, the next four sound different, then the rest sound normal.

This pattern repeats for 1-to-6 notes.

This is very difficult to describe in words, so I've recorded a video that captures the issue: https://youtu.be/pg5WvAf1TeY

It's like the voices 'snag' across different sides of the split keyboard.

This issue means that split mode is unusable for me. I've just emailed Arturia support, but I thought I'd post it here too in case someone else has experienced this issue?

Thanks.

[davydark]

The plot thickens!

After trying to isolate the issue, it turns out it affects all three timbrality modes, not just Split. That 'pop' has something to do with the filter cutoff (affects both steiner and ladder).

Working it through with Arturia support (who are being great).

Will post any update here in case someone else experiences this.

[DrJustice]

I know that when you have a big difference in frequency and with resonance on, the filters take a tiny amount of time to settle when the frequency is changed. With really high resonance you may get a sweep and a warble as the filter feedback settles. When voices are allocated dynamically, e.g. one that goes from a high frequency with little resonance to a lower frequency with more resonance will naturally exhibit a quick sweep. As each voice in your example gets a new (resonant?) lower frequency, it will 'quack' slightly (quick sweep), then the next time that same voice is triggered (as they're being cycled through) it doesn't have to do the sudden big change in frequency and so there's no 'snag' as you call it. I think you'd hear this on any analogue synth with dynamic voice allocation for different timbres.

I'm going to refrain from speculations about the implementation and how/if the effects of that sudden change of cutoff frequency can be lessened (I'm very interested in synth tech...) - hopefully you'll share the findings with us  here

[davydark]

That's an educational and informative response - thank you!

I will report back once I hear from Arturia (ticket's been elevate to 'level 2' support now, who seem pretty busy so may take a little while).

[davydark]

Just to give an update: after many emails between the retailer (not so helpful) and Arturia (amazingly helpful), I got a refund for the Polybrute because the issue affects all Polybrutes and will hopefully be fixed in a future firmware update.

[DrJustice]

I'm curious; what did Arturia say about this? As mentioned, I'd expect it to affect all analogue synths with a dynamic timbre change per voice, but I'm unsure what a firmware fix would be apart from setting new filter parameters early (earlier than now - if at all possible), gaining a few milliseconds for the filters to settle, or alternatively giving up dynamic allocation and allow a static voice reservation for the upper and lower parts as an alternative voice allocation scheme for splits.