I agree with the fact that the non-Drum voices do lots of pitch math already, so it likely won't have an impact... But I'm still going to try for the Drum Preset to see if the result gets better or stays the same. I hope it will not make a difference.
Had not tried Tabla yet, but I had a feeling that I am getting too much crashes, but was not sure if it is custom presets that causes that.
Hypothesis with sample rate is interesting, at least most of my custom presets use existing factory presets as a template, and not a lot there could be done wrong, so there is also not lot of other things we could try besides sample rate regarding crashes.
But to my knowing there are factory presets with non-48khz samples, so probably Artiphon decided that it is fine at some point. And lead presets are doing a lot of pitch shifting anyway for non-drum instruments to fill the gaps between the notes which have their own dedicated sample, not to mention pitch bends.
P.S. hope that it is not a cause, as changing sample rates will most probably ruin frame-based loop markers in the SF2 converted stuff I do.
It's still crashing a little. I'm going to convert all samples to 48Khz and change the pitches to -1 as using -1.08844 likely is causing a little more stress on the tiny little CPU.. I'll report back if it actually makes a difference.
For anyone that has tried Tabla Drum Preset, do you find it is crashing the Orba too often? I found this was happening with the PanDrum too when I was developing it. I found that the template I was using contained:
<ModifierChain chainIndex="0"> <Compatibility majorVersion="1" minorVersion="0"/> </ModifierChain>
Even though the ModifierChain was not used, somehow this must instantiate a modifier chain (which likely consumes memory and chained events). I removed this from PanDrum but forgot to for Tabla. I've made this change locally and it seems to perform much better.
@Ignis32 I just tried PocketMIDI!!! Wow what a great tool to have for debugging! Thank-you!
Steps to Install:
Here are the Tabla presets (one Drum and one Lead). They both use the same SamplePool and are bundled together. I made 8 levels of velocity for 5 Tabla sounds and repeated some of the sounds for easier playability. I still don't know why the Lead is quieter. I'm also not convinced its possible with the Orba to get a wide range of velocities (like a MIDI keyboard). I think the Bezier Curves have something to do with the range of velocity inputs it can interpret. I started looking into this with Orba1 and quit but might look into it again...
In PocketMIDI app you can set velocity manually when sending them by USB to the Orba2 from the virtual keyboard.
Used that for testing velocity thresholds, as it is much more reproduceable than trying to beat out the specific velocity by hand. PocketMIDI exists for Mac as well.
Also in your case you also could write a python script for that to run automated testing :)
P.S. Btw, seems to me that Orba2 gives much higher velocity response when being held in hands, than when it stands on the table. Probably that's due to velocity being calculated from accelerometer.
I'll post it this weekend. I'm not convinced the velocity thresholds are functioning properly as a Drum Preset. I'm going to re-create it as a Lead to see if it works better.
I found some samples! I'm pretty happy with what I have so far:
https://www.youtube.com/watch?v=3R_x80VVC60
Chris jones
1 person likes this idea