Orchestrating with modeled instruments

Another quick update for those following my modeled instrument fumblings.

Leap Motion is THE controller for bowing the SWAM Strings!

And as previously speculated, it does indeed make all the difference in the world for the realism and expressivity of these instruments. Being able to directly control bow changes vs slurs, along with bow weight and contact point (sul pont - sul tasto) takes things to a level you’re not realistically going to achieve with samples.

Of course, it takes some practicing to make it work, though, and given that I cannot for the life of me get it right without using my actual bow hand, I suspect it’s even harder if you don’t actually play bowed strings. (So, I have to move the Leap Motion over to the right side, and “finger” the keys with the left hand. Which is probably a good thing either way, since one is supposed to play keyboard with both hands anyway. :wink: )

Issues with SWAM Strings:

  • The “bow” is too short! I can of course scale the Leap Motion input, but it’s still like playing with a 12" bow; only… magnified. It’s something with the speed/dynamics mapping.
  • The bowing mode feels filtered and sluggish, which makes it tricky to play distinct articulations in sync with the left hand. Probably a result of it trying to clean up the 7 bit MIDI CC input. 14 bit would have been VERY useful. (Leap Motion supports that.)
  • The GUIs are very resource intensive or something. They’ll pretty much freeze the entire Cubase “desktop” while playing with 8 violins open, so those have to be closed whenever not actually tweaking stuff. No big deal, but a bit annoying… (This is the old GUI, though. Haven’t tried that with the updated Brass instruments.)
1 Like

That’s great news! How do you control slurs, legato times etc using the Leap controller?

PS. I hate the 7-bit resolution for MIDI 1.0, so I am eagerly awaiting full adoption for MIDI 2.0.

1 Like

Just like with brass in breath controller mode, the keys basically just select pitch without creating sound, so I just use key velocity for legato slide time, which allows for simulation of shifts to some extent. Going to have to tweak the range and curves there a bit, as nothing much happens until practically unplayable low velocities, but the basic idea of transferring ALL dynamics control to the expressive controller, and using velocity for slur/legato speed seems logical.

Only missing part here is controlling the string crossings, but I’m not sure how to go about controlling that in real time, as it doesn’t really map to the keyboard, where you can’t even see/feel where the open strings are.

Oh, and explicitly playing open strings would be useful as well, as that’s a pretty common technique in solo repertoire, due to the distinctively different tone, but I’m not actually sure if SWAM Strings implement that properly.

As for the Leap Motion, I’m currently using L/R for bowing, U/D for bow weight, and B/F for contact point. Also using the same roll/yaw vibrato depth/rate controls as for brass, but that kind of interferes with the bowing, and vibrato speed is also “backwards” as I’m using the right hand, so I’ll probably use something else for that.

Now, I’m wondering if it’s possible to have SWAM Strings respond to expression as if it was literally bow speed instead of bow position, while still behaving as in bowing mode… That way, I could circumvent the 7 bit resolution, and the bow length restriction, by doing the position-to-speed transform on the controller side.

1 Like

Agreed what people expect for not much money.

Having had a LEAPMotion for a while now, I find myself using it only for a few things because of some of the things you touched on - the moment of indeterminacy when it first senses the hand, the jumpy response (in many lighting situations), and the latency to me is a great annoyance. My TEControl is much more responsive. I will add that I use the Roland joystick on my RD700GX for pitchbend and that I don’t use any of them for doing vibrato - only on guitar instruments sometimes, and only with the joystick, but that’s what I’m comfortable with. (As an aside I feel like pitchbend wheels ought to require a license to use. There should be tests before their use is allowed. :grin:) In vibrato land I am willing to use the vibrato in, say, the SWAM Oboe, with the caveat that bite (on the TEControl) does intensity and tilt does speed. It constrains the vibrato to a nice consistency but allows the expressiveness of slowing or speeding it up. I’ve been eyeing those ring controllers for their speed (ironic that they seem to be faster than a dedicated usb device like LEAP, but all they have to do is sense their own position, not interpret a hand), and the Touché or whatever that thing is called, because it’s tactile in a way that my iPad with ComposerTools Pro isn’t.

Also - anybody else doing this? I use GeoShred Pro with a MIDI out only program - it lets you define scales or chords easily, including being able to automatically change the scale when you are moving upward or downward - and primarily I use it with harp, which is awesome. No more doing pedaling with MIDI or using prerecorded glisses - I just play what I want. ComposerTools Pro has a harp feature also but there is an elegance to how GeoShred approaches it. (CTP constrains the scale as a harp would but you don’t need to do that in GeoShred, though you can if you want to.)


And yeah, those SWAM strings are a mixed bag. Very resource-intensive. I have a template that has them on another 12-core Mac Pro so it doesn’t turn my main machine into an idiot. Even then, though I love the potential and how responsive they are, they were a bit cumbersome - and even with something like Divisimate handling some humanizing or tons of tweaking in Cubase, I wound up shelving it and using SampleModeling strings. (Using their ensembles when that’s what I want.) But maybe SWAM is better if you want that bowing thing. Perhaps something that supports OSC for better resolution? or maybe we all just have to wait for MIDI 2.0. And buy all-new controllers. :confused:


Had a quick look at Organteq while I was at it. I have some awesome sounding sampled organs that I’ve been meaning to use, but they never really inspired me to actually do something. Organteq instantly gives a completely different impression! Can’t quite put my finger on what specifically makes it feel so different, but it’s basically the same thing as modelled instruments in general: It feels like you’re playing an instrument, rather than triggering samples.

Note-off handling in sampled organs tends to be uninspiring at best (only so much you can do without introducing obvious artifacts), and you can’t really do much with stops and other “advanced” features, beyond selecting a “baked” preset and play it. (A big organ just has too many stops and other controls to be fully deep-sampled.) Meanwhile, Organteq stops respond in real time, and there’s is even an “expression” pedal that lets you scroll through a range of presets while playing, for amazing swells.

In short, this one doesn’t make me want to slap some organ parts onto the orchestrations. It makes me want to compose specifically for the organ.

1 Like

Oh, and as for the SWAM Strings bowing issue, it does actually seem like the different input modes for the bowing articulations are actually just interpreting the input differently. The bipolar mode seems to do exactly what I want; it takes input as directional bow speed, so an external processor should be able to derive that from position data from a long, high resolution ribbon controller or something.

Or, I build something with a roller that senses rotational speed, and also a force sensor in each end, and just use a bow with bass rosin on that, to generate speed, weight, and contact point. That has the bonus of responding correctly to the bow being off the string (that is, not at all), so you can bow properly, rather than being forced to rebow “in place” whether you want to or not.

1 Like

Not much “actual” music work done today (and on that note, I’m not sure about this month’s competition, as there’s just too much else going on) - but as a hint as to the general direction in which this is going, I may or may not have installed Visual Studio and JUCE… :slight_smile:

1 Like

How about some actual sound? :slight_smile: Just got my SWAM strings ensemble to a usable state, and decided to try out some stuff. There are timed comments explaining which instruments are used where.

From the description:

Working on an all modeled orchestra template, and did a quick test of the strings section. The strings ensemble is 10-8-6-6-6; seating as shown in the picture (from Precedence); that is, 36 individual modeled solo instruments. Each instrument feeds a bone dry mono signal into its own linked Precedence/Breeze pair with a dry “studio” preset for spatial positioning, followed by a send to a strings-specific IR from So. Cal. Hall, followed by a TC VSS3 instance to add some motion and smooth out the tail.

I recorded a simple 8 bar arrangement with one part for each instrument, and then arranged that into a “piece” by just using different permutations of these parts, with an octave transposition in a few cases. It was all recorded at 150 bpm, but then I play around with the tempo map to see what happens when the performances are “stretched.”

The Leap Motion is set up so that I can basically bow the instruments like a real violin, with weight and contact point on the height and depth axis respectively. Due to technical limitations, the “virtual bow” is way too short, and simultaneously rather insensitive, but with some training, it work surprisingly well as long as notes are either quiet or short. Unfortunately, one can’t edit out undesired rebows with this control method either, so I’m probably going to have to come up with a custom solution that translates the bowing motions into velocity instead, so I can use the bipolar bowing mode in SWAM.

Cubase Pro 10.5
GECO 1.3.0
SWAM Solo Strings 2.1.0
Precedence 1.5
Breeze 2.5
Spaces II
Ozone 9

Leap Motion

1 Like

Found another option for (literally) adding more expression to the traditional keyboard:

It basically turns each key into a multitouch X/Y pad. It doesn’t seem to literally sense velocity or pressure, but it does sense contact area, which sort of covers pressure.

1 Like

Oh cool, it seems this area (expressive control) is gaining traction! :smiley:

1 Like

Quick update. Haven’t been feeling well lately, and my NaSoAlMo challenge is not going very well so far - but I’ve learned a few more things in the process:

  • I rather like the tone of the saxophones in Infinite Woodwinds 2.0, but now that I’m used to playing SWAM brass with full breath controller articulation, it feels a bit awkward going back to triggering notes with the keyboard. Setting Attack Range to 0 almost fixes that. (I’m going to hack some MIDI/controller software that can translate behaviors like this, but probably no time for that right now.)

  • Kind of figured out how to play SWAM strings with the BBC2 (expression/“bow” and intonation/vibrato), Leap Motion (bow position and weight), and keyboard (pitch) - but that way I have no bow-changes, which is pretty much a deal-breaker for anything beyond slow legato and (some) short articulations. As mentioned before, the “head shake” vibrato is not optimal for violin, but it’s still a HUGE improvement over the robotic LFO vibrato.

  • “Natural” intonation and true vibrato does seem to help the SWAM strings, but it’s not quite there yet. I have some more ideas to try, though.

1 Like

I agree, and despite the fact I do sample my own stuff it’s a big time drain. Don’t get me wrong, it helps you stand out tonnes of your able to create what you need for a piece, rather than adapting someone else’s sample but there is definitely a trade off.

I tend to only make sample instruments for fun, or because no ones tried the approach I want to implement. It’s usually about the “special sauce” element of a track whenever I make a new instrument… or simply that I’m crap at playing the thing :joy::joy::joy:

Another little update, since I mentioned Infinite vs SWAM saxes, and I just grabbed the latter: YES! This is exactly what I was looking for.

Works great with the BBC2, just like the SWAM brass, and it actually “understands” both pitch bend vibrato (my BBC2 tilt trick again) and tremolo (diaphragm exercise - great for vocalists! :wink: ), so there is no need to explicitly do both tremolo and vibrato to get the intended effect. You just get more actual tremolo the tremolo way, and a more colorful retro jazz/swing vibrato with the pitch bend method. It also does that breathy ppp, essential for really intimate parts.

Tone is… different? Obviously. I don’t see a problem with it. Some much prefer Infinite, and I’d say I do to, for classical - but for jazz, SWAM sounds just right to my ears, in particular when mixed via Precedence/Breeze, which takes care of that bone dry in-your-face sound.

(Here’s a video on vibrato on the real thing, BTW: https://youtu.be/T2DTwG9Yw4s)


Interesting. Thanks for the update man, I’ll check this out as soon as I can! :smiley:

1 Like

Interesting find:

This reminds a lot of what I’ve been contemplating for controlling modeled bowed strings instruments. Didn’t occur to me simply use ribbons for BOTH strings and bow, though.

Then again, my original idea of using actual strings with force sensors on both ends has the advantage of - well, obviously - actually feeling like strings.

And, I’ve been thinking about using an actual bow on some sort of rollers or other sensors. The bow is an instrument in itself (for better or worse), and for realistic articulations, the logical approach would of course be to use an actual bow, rather than trying to simulate everything going on in it.

1 Like

Professional violinist perspective on SWAM:

Oh, and SWAM Solo Strings 3.0 is coming out this month!

1 Like

Nice! I am looking forward to the update. I hope they will also make their woodwinds use their new GUI and features later this year. Even if they are not perfect, they are the leaders in physical modeling right now! :slight_smile:

1 Like

Yes, looking forward to both upgrades (I have the full set now) - and yeah, they certainly have their issues, but I think there is a lot of confusion as to where the problems actually are. It’s easy to get the impression that it’s the actual sound that’s problematic, when it’s actually a matter of control and expression. As I’ve mentioned before, real bowed strings can sound extremely synthetic if you manage to play them with absolutely no expression at all (which is a tough technical challenge!), very much like what SWAM Solo Strings can sound like at times, but that is a sound one does not normally get to hear from real instruments, for obvious reasons.

1 Like

It’s basically similar to singing a perfect note…meaning without any vibrato or fluctuations. Extremely hard to do both physically, and mentally because we WANT to add expression :stuck_out_tongue:

1 Like