Artiphon

I still don't understand how it all ties together but I believe MPE X & MPE Y refer to the X/Y movement on each pad. Something like this:

So I hooked up PitchBend to MPE X on Vibrato Bass 2 like this:

<SeekerEntry inputLength="Default" inMin="-0.025" inMax="0.025" outMin="8021" outMax="8363" seekerType="Controller" controller="Pitch Bend" triggerSource="Default Note Source" triggerRule="Source" metricSensor="Default Note Source" metricSelection="MPE X"
 maxRate="20"/>

Refer to inMin & inMax which define the hardware In range. When you put your finger down, you define 0. So when you start in the middle and go left/right you would be in the range [-1,1] which you see in plenty of Presets. Here I changed the range to [-0.025, 0.025] so that you immediately hit the min/max on each side. the min and max correspond to outMin and outMax which are 8021 and 8363. These are equidistant from 8192 which is considered the middle value for PitchBend (a.k.a. 0 PitchBend). Those numbers yield a semitone in each direction but the range could be increased. I can imagine an instrument where you wiggle your finger back and forth and it snaps between the root note and a maj/min third up and back. This would be the same technique as how old computers (e.g. Commodore 64) used to fake chords by alternating between various notes very quickly.  

@DavidBenton

  "I suppose it's now a case of finding if there is a way to catch the tilt data as it's happening"

You can record tilt for other sounds.. So it must be it records the controller messages that it is mapped to. Many sounds map to to a filter so that you can sweep a sound while playing. Maybe the Orba is simply recording the controller events and not the tilt data itself. Now that I think of it, that is the case because I decoded Base64 song data last year and found that to be the case. Pitch Bend requires 14 bits of data instead of 8 bits like normal controllers. Maybe this is why it doesn't record pitch bend. I'll check on Orba1 but I believe it did record then. 

While trying to get to the Hazelwood chord preset I posted above,  two other presets appeared that I didn't want to delete.

The first is AlmondTree, which was the wrong sound for what I wanted - but I liked anyway. Cobwood however, was simply wrong - but the right sound. Again, I liked it with it's faults. You'll see if you try it, but you might just like it too! (I really do like flattened sevenths!)

@subskybox - OK I know I am probably insane for even imagining this would work, but I tried taking your Vibrato bass preset and adding the relevant seeker  from the other one (changing MPE Z to X) and found I could get both semitones up and down on the pads. It only worked when the finger was moved, and it was rather fussy about it, not entirely reliable, but it worked. I still don't quite understand properly why either of those presets work - but the result of the combination was sort of cool.

We could really do with having an Orbasynth app for the Orba 2 - using the current version is very hit and miss. Apart from the problem that sounds on a computer are not the same as on the Orba, I find Orbasynth on the Orba 2 can often behave very erratically and becomes useless (other times it's OK) - but not ideal. 

Also I have noticed that the factory presets mostly use the same set of seekers - so I have resorted to sticking to them until I need to do something different.

I did get round to having another try with your bezier curves but with no success - I'll have to take a look at your presets later.

Glad you liked those presets - your tilt/snap is amazing. I've not really got into recording stuff but that's obviously a big downside to it - but at least it enables better noodling about and playing along with other things. I suppose it's now a case of finding if there is a way to catch the tilt data as it's happening, and then finding if it can then be sent to the right place externally.

Orba2Bass2 Tilt/Snap demo:

https://www.youtube.com/watch?v=13dGpcaESps

Those are great sounds/chords! Nice work.

I've been working on a few altered Factory Presets that have snapping semitones. Not quite 100% happy with them but they have potential.

First is Vibrato Bass 2. It is the same as Vibrato Bass but I messed with the Bezier curves and Seeker ranges so that it will snap up or down a semitone when using vibrato. https://www.youtube.com/shorts/pSW1O1z9U9g

The second is Orba2 Bass 2. It is the same as Orba2 Bass but it will snap up a semitone when you tilt past half-way. 

These techniques can be fun to play but they cannot be recorded as performed :( 

Two synth based chord presets here:

Hazelwood - an attempt to create a preset that works like an Appalachian (mountain) Dulcimer with two melody strings in unison and two drone strings. Traditionally used for accompanying songs and tunes. Each key will sound the drones at the same pitch but you can get a melody as well!

MonksFifths - a rather monastic sort of sound with parallel fifths all the way.

(Both of these use flattened 7ths in the major - because I think it suits them better.)

Yes, the noteThreshold thing I got sorted now. The sampleGroups I should be able to sort out OK it's mostly a problem of concentration! I just about understand it now. 

oh. by sampleGroup  i've meant sampleMaps here.

goddamn uneditable posts.

@Ignis32 - Yeah, I'm afraid I'm an aged dinosaur who's so used to going to notepad that I tend to ignore anything else. I have notepad++, and you're right, I should use it! As well as the other things. (Proof reading is a weakness for me.)

I have started creating a set of all the Orba2 presets, partly for myself, but certainly for others as well. At the moment I'm grouping it by the authors. It would be useful to have some concensus about this, and also views on our experiments. I have a couple that work OK, but are not what I was aiming for and I consider them incomplete or work in progress - but possibly others might like them, or find them useful. Some feedback on these things might be useful. I don't mind getting it together but I don't want to include things that people feel are not fit for distribution for whatever reason.

Since I don't have an Orba 1 I'm not inclined to do similarly for those.

@DavidBenton

Couple of additional things:

1) there is a dot instead of comma

2) Actually, 79  even should not be here. It is not obvious, I would even say it is the opposite to obvious, but in noteThresholds  always has one element less than  amount of sample groups.  Orba silently assumes note range to 127 for the last sample group.  

If you end notethresholds with 78,  there will be an "invisible" additional range from 79 to 127 . ( You do not write 79, as everything after 78 will use the same sample group anyway for all the high notes)

Think of it like there is  always an invisible 127 in the end of noteThresholds:

-------------------

Let me write a simple example to illustrate:

With samplegroups   [0][1][2][3]

You have noteThresholds like this -  50,60,70

it will mean, that :  

sample 0 is used for (0-50)

sample 1 is used for (51-60)

sample 2 is used for (61-70)

and sample 3 is used for (71-127)  -  and there should be not record in notethresholds  for this range, it is assumed  silently.

@DavidBenton

>> I understand it Audacity converts 16 bit files to 32 bit on loading them - 

I got some into the same pitfall with Audacity messing with file format and bits silently.   In my case it was even not the correct pcm format  generated by Audacity for some reason, but something else. Turns sound into completely crazy  crushed sounds.  Currently, I use Goldwave to verify wav format, but it is not free and works in a trial mode, so if you guys will find reliable option to ensure the wav format - please share. For me it is surprisingly hard to ensure the exact format, while it should be really something basic.

>> I'm really confused about sample groups 

For me, the key to understand sample groups was to notice that note thresholds is always one element shorter than velocity thresholds and sample groups. 

It is described earlier in the thread in details, but TLDR it works the following way -

from noteThresholds orba understands which sampleGroup will be used, and then chooses sample from the group using corresponding velocityGroup. 

If you are confused with a fence of [][][]  stuff and wonder what else can be written inside - download  Subskybox's   Pandrum preset (I do not see it in this thread, so I am attaching it here). This preset is a very useful example to understand how notethresholds/velocitythresholds/samplegroups work together. At least that's how I learned it. From the  Pandrum.

>> There were a couple of errors in my Dulcihack preset posted earlier

Which editor are you using?  You have error in this version as well, sorry)

If you are not using editor with syntax highlight, I would highly recommend start using VSCode or Notepad++, so you would be able to spot issues like that visually much easier.

Also it is a good habit to check XML in online validator, like https://codebeautify.org/xmlvalidator  ( there are a lot of onlinr beautifiers/validators out there that you can like more)  It could tell you if your XML is wrong from the syntax perspective. 

P.S.

VScode  has free extension that can detect malformed xml while editing file.

I had not tried it before, (while I should)  but right now installed it and it was able to detect and show malformed part of your xml, so you might give it a try:

@Subskybox

>> thinking that we should combine all of the posted Presets 

It is a long shot, but orba storage space is quite small. Repository/store could be a better solution in the future,  with ability to choose what to install. :D  Just dreaming though.