I built out a nice bank of midicake patterns that I can jump between, so the way I cue this live is to press LOAD on midicake and leave that page open so when I notice other transition cues musically I can easily glance over and one press launch next pattern synced with ie: drum machine change
I've noticed a lot of strangeness with this where sometimes the downbeat gets slightly shifted back or forward where midicake's downbeat is off +/-3/16 The weird part is that if I give a long cue to midicake ie: press the next pattern on beat 2 so that it has a few beats to count it starts to lose things.
However, if I time it and cue right on the and of 4 it seems to land together way more consistently. I've also gotten better at a live reset stopping on 3 and restarting on downbeat, but tapping early is far preferable
Some Notes: if I run at 133 or so I need to do lots of manual resets and midicake seems to really only land if I nail the cue on the and of 4
lower bpms had less of an issue ~111 or so, however. If I had previously been at high tempos and getting lots of drift then it doesn't matter what tempo I use and drift expands.
If I do a full power disconnect and plug back in, things sync up better for a bit.
This makes me think maybe there's a memory leak or some traffic getting clogged? I was also sending new chords to midicake to coincide with the program changes and it feels like midicake ever so briefly pauses to process this and then comes in late
Ok, I have it set to retrigger right now wanting new patterns to start on 1, one follow up, would -16 beat sequences have any affect on things? I thought maybe if it was going by end of current sequence as opposed to the bar that might affect things or increase computation load.
There is also a regular flux in tempo when following on beat 4 and it feels maybe like I'm slamming a lot of data into it right when it's computing that and the program change and it briefly loses the downbeat. I also setup the arp as a bit of a hub for the moment for usb/analog links with midi thru which I recall saying it was a software implementation.
Is there a preferred input for tempo following that arp is best in? I thought if overloading was the issue it might help to have some menu config defining inputs similar to outputs. Something like USB -> CLK ONLY. I don't know if this would help things in the end, just a thought :)
I'll try cueing the chords earlier, I need to think about how to organize that for live things
Firstly, Do you have the Patch Change Behaviour to Non-Stop or Retrigger?
Also, patch change switches on the bar. You'll see it says "loading..." before switching. ARP does need to recalculate the 4 sequences when loading in a patch. However, this should all be done within a single cycle, well within a MIDI Clock tick. There shouldn't be a pause when calculating as the MIDI output has priority. However, if you are sending the input chord directly on the beat, ARP will calculate the new sequences and apply them on the next active step (usually within 1 clock tick). So ideally, you want to be sending the new chord before the next beat you want it to play on. If you can share a video of the behaviour it will help me understand the setup and shed some light on the steps to recreate if there is an issue.