When is complicated/complex unacceptable?

[Tod]

I created a Pedal Steel Guitar instrument in Kontakt some 6 years ago. I've used that instrument many times on my
various clients projects, some are country & some are not, it worked very well. I've often thought about offering it
to the public, either with a small charge or even free. But it took over 25 controllers to make it sound like a real
Steel Guitar, along with 5 keyswitchs that determined not only which controller to use for slides, but how much the
note would slide going from 0 to 64, 64 to 127 (64 is on key).

Right now I'm working on my 2nd Steel Guitar library and I'm kind of in the same situation because, the steel guitar
is such a complicated/complex instrument. My plan would be to have at least 2 separate nki instruments, one for
performance and one for programming. I haven't really thought about the performance nki yet, but I do know there's
no way that a keyboard can play all the nuances of a real steel guitar, it takes some serious midi programming to do
that.

At this point, I've actually got a good professional steel guitar sitting in my control room so I can record the samples
at will, and I've learned a great deal about different ways the steel can be played and sampled.

Here's a short video I put on YouTube a few weeks ago. what I did is DL a video of a steel guitar player playing
"Crazy Arms", then program a backing track along with my steel to replace the original soundtrack.



My situation is that as much as I love my new steel guitar, it's even more complicated then my first one, but wait,
there's even more to this. I'm a Reaper user and because of that I've been able to create my own toolbars to make
my steel much more usable.

I won't go into the details of how this all works, but I will say, that with these two toobars, I can fairly easily program
my steel guitar in Reaper's Midi Editor.

So my question is, when is complicated/complex unacceptable?

Click to read more...

 

[Reid Rosefelt]

This is hard to do on a keyboard without undue complexity, so your library may be a specialty thing. But there are lots of new, more expressive controllers that could make it much easier and intuitive to play something like this. Why not program it using the MPE spec, so that it could be played on an instrument like a ROLI, or better yet on a Linnstrument?

 

[Tod]

Wow Tiger, you've sparked my curiosity. Ha ha, I've still got the original LinnDrums laying around here somewhere that I bought back in the early 80s. Also I'd never met or seen Roger Linn before so it was very enjoyable, he is truly one of a kind, and he's still motivatin'.

However, his new controller basically works with solo instruments and although it is amazing, controlling 2 or 3 notes all going in different directions or amounts, is a little more complicated. I googled ROLI, but didn't check out any YouTubes, I'll do that when I get the chance.
Thanks Tiger, I wasn't aware of these things.

 

[Reid Rosefelt]

These articles might interest you...

https://reverb.com/news/mpe-the-midi-killer-and-future-of-musicianship-in-electronic-music
(broken link removed)
(broken link removed)
(broken link removed)

 

[Tod]

Indeed, very interesting. It would also be interesting to look under the hood, basically they are assigning multiple midi controllers to one physical controller, which is exactly what I'd like to do with the steel.

Roger's design appears to use multi channels, however I didn't see any mention of whether it's midi channels, nor how many channels. I assume though it would be midi channels, possibly 16 unless they in some way use ports.

Keith McMillen's controllers seem to be the most feasible right now, and not priced badly at all.

Thanks again Tiger, you brought me back into the 21st century and I appreciate it.

However, heh heh, I'm still stuck with my same question about "when is complicated/complex unacceptable"? Maybe another way to put it, "when you are programming midi, how many controllers are too many"?

 

[Parsifal666]

When it's way too obvious that's exactly what you're trying to do.

 

[Tod]

When it's way too obvious that's exactly what you're trying to do.

Ha ha, thanks Parsifal666, is that what you think I'm trying to do?

 

[Parsifal666]

Actually it seems you're doin' pretty good there, my friend. I just winged it out of context, so take it for the little it is.

 

[richardt4520]

If I'm reading Parsifal correctly (I think I am), he's right. Your problem is your scope is too broad. You're attempting to make basically a virtual instrument that will do everything the real instrument can. That's going to be way more complicated than most people will want to deal with. As a guitarist, there is no library out there that can do everything that a competent guitarist can do in its entirety. heck, I don't know many guitarists (I certainly can't) who can play every style very well themselves. But there are lots of great guitar libraries out there that can pull off something that sounds pretty realistic. And it's because they focus on a few things instead of trying to do everything the real instrument is capable of. They would fail at that because it would be a tool that would be way too complicated for what the majority of people want-just something that can produce a believable guitar part quickly, without much trouble.

 

[richardt4520]

So basically I've been no help at all other than to say dumb it down for us lazy people! haha

 

[Reid Rosefelt]

Indeed, very interesting. It would also be interesting to look under the hood, basically they are assigning multiple midi controllers to one physical controller, which is exactly what I'd like to do with the steel.

Roger's design appears to use multi channels, however I didn't see any mention of whether it's midi channels, nor how many channels. I assume though it would be midi channels, possibly 16 unless they in some way use ports.

Keith McMillen's controllers seem to be the most feasible right now, and not priced badly at all.

Thanks again Tiger, you brought me back into the 21st century and I appreciate it.

However, heh heh, I'm still stuck with my same question about "when is complicated/complex unacceptable"? Maybe another way to put it, "when you are programming midi, how many controllers are too many"?

MPE is a new spec within MIDI
It sends out data to various tracks at the same time. Starting with channel 2 and then going up. Each of your strings is on separate MIDI tracks. So, for example, you could have one string bending up, one bending down and one not bending at all, etc.
And when you pitch bend (or slide) it isn't just one wheel bending everything, you can slide some strings up and others down. You can get a ROLI Seaboard Block for $300 and send it back after a month. But there are tons of videos on YouTube.

The video below isn't true MPE, but something ROLI software calls multi-mode, but it should give you an idea



Recently Orange Tree samples have updated their instruments to have a new feature called "polybend," which the capability for polyphonic pitch bends within a single instance of a library.

Keith McMillen's controller is the most like a piano and therefore the easiest to learn how to play, but it remains to be seen if it will be the most expressive. It may be necessary to take the time to learn one of the other MPE controllers to get the results you want. The ROLI controllers are very cool and are much cheaper than older instruments like the Continuum.

Make an instrument for MPE is a little future-oriented today. But in a few years time, I believe there will be MPE controllers from Roland, Yamaha, Arturia, Alesis, AKAI, etc. There will be a lot of different kinds of controllers out there at a lot of price points, and more and more people will be experimenting with this stuff. Particularly anybody who is trying to capture the sound of an acoustic instrument that isn't fretted or keyed in whole tones, and can bend or slide up or down, like a violin or a harmonica or a sitar or... a pedal steel guitar. And that certainly means the people who habituate this forum. After all, who used the Continuum Fingerboard? John Williams, Ramin Djawadi, AR Rahman, etc.



The reason I'm telling you all this is that these new controllers may make it more intuitive for people to play VIs in a realistic way without manipulating a bank of sliders, and learning a lot of keyswitches. But all this is still in its infancy.

 

[ohernie]

When the level of complexity - controllers, whatever - approaches the level of complexity of playing the original instrument, then it's too many. The only possible advantage at some point is price and convenience, as in setup time and editability.

The sound is great, but what's the advantage of learning to play an instrument with 25 controllers versus learning to play the original?

 

[Parsifal666]

When the level of complexity - controllers, whatever - approaches the level of complexity of playing the original instrument, then it's too many. The only possible advantage at some point is price and convenience, as in setup time and editability.

The sound is great, but what's the advantage of learning to play an instrument with 25 controllers versus learning to play the original?

There are instances of this. Dumbing down is a good choice when you're looking for actual $. And today that goes more than ever for music in general imo.

 

[ohernie]

Yeah, this gets back to the eternal "how realistic should a VI be" thread. The whole cycle started out with people wanting Garritan because it was more realistic than a Sound Canvas. Then developers started flying too close to the holy grail - VIs that could actually substitute for acoustic instruments in paid situations - and messed everything up <g>.

Because of the complexity of the original controller this project becomes two engineering projects, the sound and the controller, and it appears the sound is the easier of the two. Actually, this project might require a third element - a set of steps that define a recording protocol.

As with other VIs, 25 controllers is OK IF they only need to be tweaked during editing. The real question is: What is the minimal number controllers needed to lay down a more or less realistic track? If the player can lay down a track with the correct feel, they can always go back and finesse it.

 

[storyteller]

I have long thought about how this should work since I lived in Nashville, but I honestly have no desire to sample a steel and program it. But here is the actual solution:

You have two sets of key groups on the keyboard. The "right hand" key group has 10 primary keys. These represent the 10 strings on the pedal steel in ascending order as they are viewed from the player perspective. Don't pay attention to note names yet.

The "left hand" key group represents each fret on the pedal steel.

The right hand plays the "picking patterns" on the 10 strings. Low velocity is palm mute. Hard velocity is sustain.

The left hand represents the bar placement on the fretboard. Overlapping notes represent a legato slide transition for sustained notes.Velocity of overlapping notes represents the duration of the slide.

Yes, the bar on a real pedal steel can be placed diagonally across multiple frets. The solution here is that two keys played simultaneously with the left hand represent each end point of the bar on the fretboard. Some crafty programming and math regarding bar-size versus start and end points will give you the precise sample and intonation of each key played. Of course, a legato transition for the bar will also need to be accounted for. This is by far the most tricky part of the programming. The rest is cake.

Simple on/off controllers represent each foot pedal on the pedal steel. Customers could use any midi foot pedal with four or more pedals to play this in realtime.

Bar vibrato: Aftertouch for realtime or CC1 for more precise editing.

Volume swells? Expression pedal. This is the same as how a real steel player would use a volume pedal.

Now the debatable part. The knee levers. Depending on implementation, you could take one of two paths. You could go with another set of foot controllers per lever, or one additional set of keys in between the main keys to trigger half and whole bends up and down. Velocity would trigger that. I think I prefer the first method to keep the implementation as a perfect representation of the instrument. But again, it would depend on playability. Everything described up unto the knee levers would be a perfect 1-for-1 implementation.

A roli seaboard would offer additional options, but that would also limit your customer base. In the end, you would need to make sure your programming is tight on this. However, Kontakt can easily handle everything described above.

Enjoy!

 

[pmcrockett]

I think the existence of things such as the Audio Modeling strings and EWQL Symphonic/Hollywood Choirs demonstrates that you can pretty much make things as complicated as you want. But as with those two examples, you should make sure that users who don't want to deal with the complexity can still get serviceable results without engaging with all of the complexity.

 

[Tod]

If I'm reading Parsifal correctly (I think I am), he's right. Your problem is your scope is too broad. You're attempting to make basically a virtual instrument that will do everything the real instrument can.

So basically I've been no help at all other than to say dumb it down for us lazy people! haha

I'm sorry richardt4520, I got busy and couldn't answer till now. You're telling me what you think and that's exactly what I want to hear. And you're right, I am trying to make a midi steel guitar that can sound real. I'll admit, it's mostly for myself, but I wouldn't mind being able to sell it, or maybe even give it away at some point down the road.

That's going to be way more complicated than most people will want to deal with. As a guitarist, there is no library out there that can do everything that a competent guitarist can do in its entirety. heck, I don't know many guitarists (I certainly can't) who can play every style very well themselves. But there are lots of great guitar libraries out there that can pull off something that sounds pretty realistic. And it's because they focus on a few things instead of trying to do everything the real instrument is capable of. They would fail at that because it would be a tool that would be way too complicated for what the majority of people want-just something that can produce a believable guitar part quickly, without much trouble.

I've played guitar too, nearly my whole life and I totally agree with you. It's not nearly as complicated as a steel is, but for me, the guitar has far more nuances. I've never tied to duplicate my own guitar playing, I think that would be impossible.

 

[richardt4520]

Haha! I hear ya! Just looking at a steel scares me! I don't think I'm coordinated enough to play one of those!

 

[Tod]

When the level of complexity - controllers, whatever - approaches the level of complexity of playing the original instrument, then it's too many. The only possible advantage at some point is price and convenience, as in setup time and editability.

The sound is great, but what's the advantage of learning to play an instrument with 25 controllers versus learning to play the original?

Very well said ohernie and thank you. I think learning to play a steel guitar is a little more than learning how to use my steel in Kontakt. I've recorded a half dozen or so steel guitar players over the years, a couple of them were great, the others not so. For myself I'd like to add steel to some of my clients songs and we just don't have the great steel players around anymore.

 

[Tod]

MPE is a new spec within MIDI
It sends out data to various tracks at the same time. Starting with channel 2 and then going up. Each of your strings is on separate MIDI tracks. So, for example, you could have one string bending up, one bending down and one not bending at all, etc.
And when you pitch bend (or slide) it isn't just one wheel bending everything, you can slide some strings up and others down. You can get a ROLI Seaboard Block for $300 and send it back after a month. But there are tons of videos on YouTube.

The video below isn't true MPE, but something ROLI software calls multi-mode, but it should give you an idea

Thanks again Tiger, I've learned a lot about this from you. Based on what I've learned, this would require a whole different approach for the Kontakt scripting. I would have to know everything about these controllers in order to tackle something like that.

Keith McMillen's controller is the most like a piano and therefore the easiest to learn how to play, but it remains to be seen if it will be the most expressive. It may be necessary to take the time to learn one of the other MPE controllers to get the results you want. The ROLI controllers are very cool and are much cheaper than older instruments like the Continuum.

Yeah, Keith's controllers look petty good to me, not only the price, but I'm pretty ingrained to the keyboard.

The reason I'm telling you all this is that these new controllers may make it more intuitive for people to play VIs in a realistic way without manipulating a bank of sliders, and learning a lot of keyswitches. But all this is still in its infancy.

It all makes sense, but it's going to take a while. I would have to have my hands on in order to truly know what they can do. Heh heh, we'll have to see, time is a porblem.